instrumentation for molecular electronics device … for molecular electronics device research by...
TRANSCRIPT
Instrumentation for Molecular Electronics Device Research
by
Ashley Ann Kibel
A Dissertation Presented in Partial Fulfillment of the Requirements for the Degree
Doctor of Philosophy
Approved November 2010 by the Graduate Supervisory Committee:
Stuart Lindsay, Chair
Ralph Chamberlin Thomas Moore Sefika Ozkan Otto Sankey
ARIZONA STATE UNIVERSITY
December 2010
i
ABSTRACT
This dissertation describes work on three projects concerning the design
and implementation of instrumentation used to study potential organic electronic
devices. The first section describes the conducting atomic force microscope
(CAFM) in the study of the mechanical and electronic interactions between DNA
bases and nucleosides. Previous STM data suggested that an STM tip could
recognize single base pairs through an electronic interaction after a functionalized
tip made contact with a self assembled monolayer then was retracted. The
conducting AFM was employed in order to understand the mechanical
interactions of such a system and how they were affecting electrical responses.
The results from the conducting AFM showed that the scanning probe system was
measuring multiple base-pair interactions, and thus did not have single base
resolution. Further, results showed that the conductance between a single base-
nucleoside pair is below the detection limit of a potential commercial sequencing
device.
The second section describes the modifications of a scanning probe
microscope in order to study the conductance of single organic molecules under
illumination. Modifications to the scanning probe microscope are described as
are the control and data analysis software for an experiment testing the single
molecule conductance of an organic molecule under illumination. This
instrument was then tested using a novel charge-separation molecule, which is
being considered for its potential photovoltaic properties. The experiments
ii
showed that the instrumentation is capable of detecting differences in conductance
upon laser illumination of the molecule on a transparent conductive surface.
The third section describes measurements using the illuminated CAFM, as
well as the design and construction of an illuminated mercury drop electrode
apparatus. Both instruments were tested by attempting to observe photovoltaic
behavior in a novel self-organized film of the charge-separation molecules
mentioned in the previous paragraph. Results and calculations show that the
conducting AFM is not a useful tool in the examination of these organic
photovoltaics, while the mercury drop apparatus measured photovoltaic effects in
the film. Although photovoltaic effects were measurable with the mercury drop
electrode, it was found that the film exhibited very low photon-to-electron
conversion efficiency (IPCE).
iii
For my parents, Daniel and Kathleen,
and my husband, Jonathan
iv
ACKNOWLEDGEMENTS
I must thank all those that have helped me over the years. In particular, I
wish to thank my fellow graduate students, Shreya Bhattacharyya and James
Bridgewater, who have been especially helpful with the organic photovoltaic parts
of this project. Dr. Tomoyuki Morita taught me how to use the atomic force
microscope and Dr. Iris Visoly-Fisher was a great lab teacher and mentor.
Professor Devens Gust was always willing to sit down and discuss any challenges
or results and Dr. Paul Lidell provided the molecules that were essential for my
work. Finally, I must thank my advisor Stuart Lindsay. I was financially
supported by the NSF IGERT program and an NSF NIRT grant.
v
TABLE OF CONTENTS
Page
LIST OF TABLES..………………………………………………………………ix
LIST OF FIGURES..……………………………………………………………...x
CHAPTER
1. INTRODUCTION ........................................................................................... 1
1.1. Instrumentation utilized in molecular electronic research....................... 1
1.1.1. Scanning tunneling microscopy........................................................2
1.1.2. Atomic force microscopy..................................................................7
1.1.3. Mercury drop electrodes .................................................................16
1.2. Molecular electronics............................................................................. 17
1.2.1. Energy levels of electrons in organic molecules.............................18
1.2.2 Electron transfer...............................................................................24
1.2.3. Experimental methods for molecular electronics research ..............26
2. DNA SEQUENCING BY RECOGNITION ................................................. 32
2.1. Deoxyribonucleic acid (DNA)............................................................... 34
2.2. Sequencing of DNA............................................................................... 38
2.3. Sequencing by recognition .................................................................... 41
2.3.1. Benefits of sequencing by recognition electron tunneling..............41
2.3.2. STM measurements of sequence recognition .................................44
vi
CHAPTER Page
3. ADHESION FORCE AND CONDUCTANCE MEASUREMENTS OF
DNA NUCLEOSIDES...........................................................................................51
3.1. CAFM measurements ............................................................................ 53
3.1.1. Materials and methods ....................................................................53
3.1.2. Relationship between force and setpoint current ............................58
3.1.3. Calculations of molecular conductance ..........................................67
3.2. Conclusions............................................................................................ 68
4. ORGANIC PHOTOVOLTAICS................................................................... 70
4.1. Photovoltaic energy conversion............................................................. 71
4.1.1. Semiconductor photovoltaics..........................................................72
4.2. Benefits of organic photovoltaics .......................................................... 73
4.3. Challenges to commercial organic photovoltaics .................................. 75
4.4. Organic photovoltaic architecture.......................................................... 77
4.4.1. The Gratzel electrochemical cell ....................................................78
4.4.2. Heterojunction organic photovoltaics .............................................80
4.4.2. Utilization of inorganic nanostructures...........................................81
4.5. Porphyrin based molecular dyads and triads ......................................... 84
5. ILLUMINATED STM FOR SINGLE MOLECULE CONDUCTANCE..... 88
vii
CHAPTER Page
5.1. Illuminated STM instrumentation.......................................................... 89
5.1.1. STM ................................................................................................89
5.1.2. Laser beam delivery........................................................................90
5.1.3. Environmental control ....................................................................92
5.2. Illuminated single molecule experiment design .................................... 93
5.2.1. STM breakjunction experiment ......................................................93
5.2.2. Illumination.....................................................................................94
5.2.3. Data analysis ...................................................................................95
5.3. Illuminated single molecule conductance results .................................. 98
5.3.1. Conductance of illuminated dyads..................................................98
5.3.3. Free-base porphyrin .................................................................101
5.3.4. Control experiment ..................................................................101
5.3.5. Conclusions..............................................................................102
6. PORPHYRIN-FULLERENE DYAD THIN FILM GROWTH AND
ELECTRICAL CHARACTERIZATION........................................................... 105
6.1.1. Materials and methods ..................................................................106
6.1.2. Spectral properties of dyad thin films...........................................108
6.2. Attempts at CAFM photocurrent measurements ................................. 111
6.2.1. Materials and methods ..................................................................111
viii
CHAPTER Page
6.2.2. CAFM photocurrent results and challenges..................................113
6.2.3. Calculations of expected photocurrents ........................................116
6.3. Mercury drop photocurrent measurements.......................................... 119
6.3.1. Materials and methods ..................................................................120
6.3.2. Mercury drop electrical results .....................................................125
6.3.3. Calculations of photocurrent efficiencies .....................................133
6.4. Conclusions.......................................................................................... 135
REFERENCES…………………………………………………………………138
APPENDIX
A LABVIEW CODES……………………………………………….…145
ix
LIST OF TABLES
Table Page
2.1. The genetic code. .......................................................................................... 37
3.1. Calculated conductances of base-base molecular junctions and base-
nucleoside junctions.47 .................................................................................. 69
x
LIST OF FIGURES
Figure Page
1.1 Schematic of typical STM operation. .............................................................. 4
1.2. Schematic of current to voltage converter. ..................................................... 5
1.3. Top down schematic of PZT tube scanner...................................................... 6
1.4. Schematic diagram of contact mode AFM. .................................................... 8
1.5. Graph of the Lennard-Jones potential............................................................. 9
1.6. Conducting AFM images of fluorine doped tin oxide. ................................. 13
1.7. SEM image of an AFM tip............................................................................ 15
1.8. Aviram and Ratner molecular diode. ............................................................ 17
1.9. Orbital diagrams of sigma and pi bonding.................................................... 22
1.10. Marcus energy diagram............................................................................... 25
1.11. Design and image of experiment to measure single molecule
conductance using gold nanoparticles and conducting AFM.21.....................28
1.12. Schematic of STM break junction experiment.25........................................ 29
1.13. Schematic of STM under potential control.30 ............................................. 30
2.1. Molecular structures of DNA components. .................................................. 34
2.2. Molecular structures of thymine and uracil. ................................................. 36
2.3. Autoradiograph gel for reading the sequence of DNA. ................................ 40
2.4. Schematic for a nanopore DNA sequencer utilizing molecular tunneling
recognition.39................................................................................................. 43
2.5. Schematic of STM tunnel decay experiment on DNA oligomers.45............. 46
xi
Figure Page
2.6. STM tunneling decay curves for different base functionalized tips and a
DNA-Guanadinium surface.45........................................................................47
2.7. Charge transferred during STM probe retraction.......................................... 48
3.1. Initial STM retraction curve fits for a guanine functionalized probe
retracting from a cytidine functionalized surface. .........................................51
3.2. Picture of STM/AFM experiment................................................................. 53
3.3. SEM image of AFM tip radius after metallic coating................................... 54
3.4. Procedure for obtaining cantilver spring constant. ....................................... 57
3.5. Typical force and current curve. ................................................................... 58
3.6. Plot of adhesion force versus the current at probe retraction........................ 60
3.7. Schematic of the elastic-distortion model.47 ................................................. 62
3.8. Decay curves for AFM cantilevers of different spring constants. ................ 63
3.9. Elastic-distortion model coefficient fitting.47 ............................................... 64
3.10. STM withdraw curves for G-C and A-T interactions. ................................ 65
3.11. Elastic-distortion model fits to the STM charge transferred data.47 ........... 67
4.1. Schematic of a P-N junction photovoltaic device......................................... 72
4.2. Equivalent circuit for a photovoltaic diode device. ...................................... 73
4.3. Best research cell efficiencies by type of solar cell and year.55.................... 75
4.4. Energy diagram of organic photovoltaic device. .......................................... 76
4.5. Gratzel electrochemical photocell................................................................. 79
4.6. Schematic of metal nanoparticle in an electric field. .................................... 82
xii
Figure Page
4.7. Chemical structure of porphine..................................................................... 84
4.8. Absorbance spectra of tetratolylporphyrin.................................................... 85
4.9. Chemical structure of a porphyrin-fullerene dyad molecule. ....................... 86
5.1. Schematic and picture of illuminated STM system. ..................................... 89
5.2. Conductance histograms of porphyrin fullerene dyads on ITO.................... 98
5.3. Graph of peak rations versus laser intensity. .............................................. 100
5.4. Light on and off histograms of 3-(pyridine-4-yl)-propionic acid ............... 101
5.5. Transient absorption spectra of porphyrin-fullerene dyads on ITO............ 103
6.1. Absorbance spectra of 5 uM free-base and protonated porphyrin-
fullerene dyad in 1,2,4-Trichlorobenzene....................................................109
6.2. Absorbance spectra of porphyrin-fullerene dyad on ITO........................... 111
6.3. CAFM photocurrent measurements............................................................ 113
6.4. Effect of laser illumination on cantilever bending...................................... 114
6.5. IV curves of porphyrin layer made with conductive diamond AFM tips. .. 115
6.6. Mercury drop electrode instrumentation..................................................... 120
6.7. CCD image of mercury in contact with dyad thin film sample. ................. 124
6.8. Dyad thin film photocurrent density measurement..................................... 126
6.9 Average photocurrent density versus drop position, measured from
position 1......................................................................................................127
6.10. Dyad thin film photovoltage measurement............................................... 128
xiii
Figure Page
6.11. Comparison of photocurrent action spectra of the dyad film with the
spectra of the fraction of photons absorbed by the sample. .........................130
6.12. Comparison of photocurrent action spectra. ............................................. 131
6.13. Maximum possible power density versus solar cell size. ......................... 132
6.14. Incident photon-to-electron conversion efficiency for dyad thin film
and mercury drop electrode. ........................................................................135
1
1. INTRODUCTION
This chapter first describes the instrumentation utilized in molecular
electronic research. Specifically, it examines the two main types of scanning
probe microscopes, scanning tunneling and atomic force, and quickly reviews the
use of mercury drop instrumentation. The second part of the chapter describes the
theory of molecular electronics and experimental techniques used to measure the
electronic properties of organic molecules.
1.1. Instrumentation utilized in molecular electronic research
Instrumentation is a key component in the ability to experimentally
understand the electronic properties of organic molecules. The three main types
of instrumentation that will be described in this dissertation are scanning
tunneling microscopy, atomic force microscopy and mercury drop electrodes.
Scanning tunneling microscopy and atomic force microscopy are both considered
forms of scanning probe microscopy. Scanning probe microscopy, in particular,
has paved the way for much of the experimental investigation into molecular
electronics.
Part of the increased interest in the field of nanoscience and
nanotechnology in the recent decade has been spurred on by the ability to image
and interact with structures at the nanoscale. Scanning probe microscopy (SPM)
has become a highly employed method to image and interact with nanoscale
materials. The field of SPM has grown over the last 3 decades since the invention
of the scanning tunneling microscope, and the applicability and versatility of these
2
machines has been greatly enhanced. The ability to simultaneously probe and
image a sample with the same piece of equipment, in particular, has allowed
many breakthroughs spanning a wide range of fields within nanoscience.
In terms of imaging capabilities, scanning probe microscopy is unique to
other forms of microscopy in that a sharp probe interacts with the surface being
imaged. The mechanism of interaction is different depending upon the type of
microscopy and the imaging mode, but the generated signal is almost always
highly correlated to features in the topography. This interaction produces a
measurable signal that, again, is typically related to topographical features. While
imaging, this probe is continuously modulated in order to keep this signal
constant, and the degree of modulation of the probe can be recorded and,
effectively, maps out the surface. The type of signal that is produced from the
interaction between the probe and the surface determines what type of surfaces
can be imaged and what the features in the image actually represent. The two
major categories of scanning probe microscopy are scanning tunneling
microscopy, which utilizes a tunneling current signal between probe and
conductive surface, and atomic force microscopy, which utilizes the bending of a
flexible cantilever as a surface sensor.
1.1.1. Scanning tunneling microscopy
The scanning tunneling microscope (STM) was invented in 1981 by
Binnig and Rohrer, who won the Nobel prize for the invention in 1986.1 The
STM works by incorporating a very sharp metallic probe, which, ideally, has a
single atom at its point. This tip is then approached to a conductive surface until a
predetermined tunneling current (typically in the nanoampere range) is achieved.
On a bare metallic surface, like bare gold, this tunneling current can typically be
achieved without hard contact with the surface. As is known from quantum
mechanics, the wavefunction for a particle to tunnel through a one dimensional
potential barrier is xEVmAx 2)(2exp()(
h
−−=ψ ), where E is the energy of the
particle and V is the potential of the barrier. Squaring this to get the probability
of finding a particle at any given location in the one dimensional problem gives
))(22exp()( 222 xEVmAx
h
−−=ψ , meaning that the probability of finding the
particle a distance x through a potential barrier falls off exponentially with x.
While these formulas do not as accurately reflect the more complicated situation
of an STM tip with electrons tunneling through a vacuum gap, it is clear to see the
implications. As the tip is moved closer or further away from the surface, the
current will drop off or increase exponentially. Adjusting the height of the STM
probe to keep this current constant while scanning across a surface and accurately
measuring the movement of the tip in response to these adjustments allows for
very high resolution imaging of topography.
3
Figure 1.1 Schematic of typical STM operation. Tunneling current is connected to a current to voltage preamplifier, and this voltage is used to control distance between the tip and the sample
The actual instrumentation for the typical STM measurement is fairly
simple (see Figure 1.1). An STM scanner allows for the insertion of a metal wire
tip that has typically been electrochemically etched to be very sharp.2,3
Electrically attached to this tip is a current to voltage preamplifier that converts
nanoampere currents into more easily measurable and transferable voltages. This
conversion can be obtained by simply connecting the tip to a operational amplifier
(op-amp). The current lead from the tip is connected to the inverting input of the
amplifier (see Figure 1.2). Since the other op-amp input is grounded, the
inverting input also remains at zero potential, and thus the tip is grounded. Since
no current flows through an ideal op-amp, the tunneling current runs through the
4
resistor connecting the inverting input to the output voltage. Using Ohm’s law,
then, 1RVV
i inouttunneling
−= , but since Vin
is grounded, this simplifies to
1RV
i outtunneling = , or 1RiV tunnelingout ⋅= ,
meaning that the output voltage is
proportional to the tunneling current.
Typical preamplifiers range from 0.1
nA/V to 100 nA/V, and, in many
scanners, are interchangeable
depending upon the desired current
range.
Figure 1.2. Schematic of current to voltage converter. Tunneling current is wired into the inverting input of an operational amplifier. Resulting voltage output is proportional to the tunneling current.
The tip is also in mechanical contact with an element that can control the
height of the tip, and, typically, its x and y positions as well. This element has
piezoelectric properties, which means that voltages applied to it translates in a
reliable way to changes in the elements’ dimensions. In almost all SPM scanners
this element is made from lead zirconium titanate (PZT).4 Typical PZT elements
in an STM scanner are made into tube shapes with electrodes both inside and
outside the tube (see Figure 1.3). The inside electrode is typically grounded so
that changing the voltages on all the outer electrodes to a particular voltage causes
the tube wall to either become thinner or thicker depending upon the sign of the
applied voltage. This, in turn, causes the length of the tube to either grow or 5
contract, since the volume of the tube remains constant. Applying different
voltages on the coupled sets of electrodes around the outside of the tube can cause
the tube to bend in the x and y axes, allowing for motion in the x and y directions.
Typically, samples
are manually positioned
near the tip by the user
and are then automatically
approached closer through
the use of stepper motors.
Once the setpoint has been
reached, the tunnel
junction is kept stable by
modulating the z axis of
the piezo using a controller. To image a sample, either the tip can be raster
scanned with the piezo to which it is attached while controlling tip height with the
same piezo, or the sample can be raster scanned on a piezo stage while the height
of the tip adjusts. The voltage that gets applied to the piezo then gets fed into
whatever control software is being used, and converted into a topographical image
by knowing the calibration values that convert voltage to distance on the piezo.
Figure 1.3. Top down schematic of PZT tube scanner.
One particular drawback to imaging with an STM is its inability to
definitively distinguish electrical information from topographical information.
All samples used with the STM must be somewhat conductive in order for a
6
7
tunnel junction to be maintained. For samples of highly homogenous
conductivity on the nanoscale, the use of tunnel current as a determination of
topography proves very accurate. In samples of heterogeneous conductivity,
however, the distinction between a topographical feature and a feature of high or
low conductivity becomes difficult to ascertain. The lack of ability to image non-
conductive samples and the difficulty in analyzing images of heterogeneous
conductivity have been major driving forces in the increased popularity of using
the atomic force microscope.
1.1.2. Atomic force microscopy
As another form of scanning probe microscopy, the atomic force
microscopy (AFM), (invented soon after the STM in 1986 by Binnig, Quate and
Gerber)5 shares many similar attributes with the scanning tunneling microscope.
Unlike the STM, however, the AFM utilizes a flexible cantilever with a very
sharp tip attached to it in order to probe and image surfaces. The interaction
between the probe and the surface in AFM is not tunnel current like the STM, but,
as the name implies, forces of attraction or repulsion between the two. This force
is sensed by the bending of the cantilever, which is monitored by reflecting a laser
off of the backside of the cantilever and recording the position on a photodiode
detector. Any surface can interact with this cantilever, and this means that AFM
measurements are not strictly limited to conductive samples.
Figure 1.4. Schematic diagram of contact mode AFM. The bending of the cantilever is detected by a laser bounced of its back. This signal is recorded and then utilized in a feedback loop to control the contact force between the tip and the surface
One of the most basic AFM imaging modes is known as contact mode (see
Figure 1.4). In this mode, an AFM tip/cantilever is approached to the surface
until the cantilever registers a certain force, determined by a specific amount of
bending of the cantilever upon contact. The cantilevers have an intrinsic spring
constant, so bending of the cantilever directly converts into a force of contact. As
discussed above, this bending is measured by recording the displacement of a
laser beam bounced off of the back of the cantilever. This force/bending is then
kept constant as the tip is scanned across the surface. When the tip hits a high or
low feature on the surface, the instantaneous force between the tip and sample
changes and is quickly corrected by applying a voltage to the piezo in mechanical
contact with the cantilever which either lowers or raises the force to a value
determined by the user. Thus, while mapping the surface, the force is kept
constant.
8
Although it is
a common imaging
mode, Atomic force
microscopy is not
limited to contact
mode operation.
Interaction forces
between tip and
surface actual change
considerably as the
tip is approached to
the surface. Figure 1.5 shows the interaction potential between the tip and the
surface. The potential curve is a graphical representation of the Lennard-Jones
potential, ⎥⎥⎦
⎤
⎢⎢⎣
⎡⎟⎠⎞
⎜⎝⎛−⎟
⎠⎞
⎜⎝⎛=
612
4)(rr
rV σσε . As the tip gets closer to the surface, the
12th power of the term, the repulsive term, dominates, but further away from the
surface the attractive power to the 6th term remains. The two different regimes
give rise to the two types of dynamic force microscopy which are very popular for
imaging in AFM.
Figure 1.5. Graph of the Lennard-Jones potential.
In dynamic force microscopy, the tip is not simply approached to the
surface until it hits with a certain force as in contact mode AFM. The most
common form of dynamic force microscopy is referred to as amplitude
9
10
modulation atomic force microscopy (AM-AFM). Unlike in contact mode, the tip
is vibrated at a frequency close to its resonance. The surface is then approached
to the tip. As the distance between the tip and the sample decreases, the
amplitude of the cantilever vibration will decrease as it interacts with the surface.
Once the amplitude reaches a setpoint, it is held constant as the tip is scanned
across the surface. The two main types of dynamic force microscopy are
intermittent contact mode and noncontact mode. In intermittent contact, the
cantilever operates in both the repulsive and attractive potential regimes, and as it
vibrates and scans across a surface, it taps down, making contact with the atoms
on the surface. For this reason, intermittent contact mode is oftentimes referred to
as “tapping” mode. Intermittent contact is achieved by vibrating a tip just below
the resonant frequency. Once the tip begins to tap down onto the surface, the
amplitude of the vibration is reduced, and is held constant by the servo system.
Changes in force are detected by changes in the amplitude, and the tip is pulled
away from or towards the surface to maintain constant amplitude.
In AM-AFM non-contact mode, used mainly for biological and other
softer samples, the cantilever operates in the attractive potential regime, achieved
by vibrating the tip just above the resonant frequency. Upon approach to the
surface, the amplitude of the tip vibration lowers due to the interaction with the
attractive potential, but it does not make physical contact as it would in tapping
mode. Typically this non-contact mode is used to image soft samples, such as
biological materials, in liquids. Although this can be accomplished by utilizing
the vibrating piezo usually used in tapping mode, this piezo has difficulty
vibrating the tip properly in a damped liquid environment. Furthermore, the piezo
vibrates more than just the cantilever and can cause acoustic effects in the
surrounding solution which can distort images. For this reason, many utilize a
magnetic cantilever and an oscillating magnetic field either above or below the tip
in order to oscillate just the cantilever and minimize interference. This is referred
to as magnetic AC mode imaging, or MAC mode, and its use has led to high
quality imaging and measurements of soft samples.6-8
Although not used as frequently, frequency modulation AFM (FM-AFM)
has been shown to improve upon the bandwidth of dynamic force AFM. In
amplitude modulation, a change in the force felt by the cantilever is followed by a
shift in the amplitude of the cantilever oscillation. This shift in amplitude,
however, is time dependent and is strongly correlated to the Q-factor of the
particular cantilever. This amplitude change can be represented by
[ ])()(cos2)( 02
022
02
00
ϑϑωωωω
−+−++=−−
tdtQ
t
tQ
t
t teAAeAAtA1.1
Where A0 is the new steady state amplitude after the change in force, ω0 is the
new resonant frequency of the cantilever, ωt is the frequency of free oscillation,
ωd is the driving frequency, and θ0 is the new phase.9 θt and At are boundary
condition fitting parameters. One can achieve a faster response time by lowering
the Q factor of the cantilever. However, lowering the Q of the cantilever, lowers
the sensitivity of the device in amplitude modulation. In FM-AFM, the tip is
pulled up and down based upon shifts detected in frequency, instead of amplitude. 11
12
In this case, a high Q factor does not increase the response time of the system, but
will increase the sensitive. For this reason FM-AFM can offer very high
resolution imaging at faster scanning speeds.9
Although not originally designed to measure electrical properties of
samples, changes to the typical instrumentation allow for various electrical
measurements to be performed using the AFM. Conducting AFM, for instance,
can be performed by coating the tip and cantilever with a conductive substance
such as gold or platinum. A voltage applied to the sample and a current to voltage
preamplifier in the scanning head allow for measurements of nanoscale currents
performed without largely affecting the control of interaction forces. Although
the STM can provide similar information, the separation of the conduction signal
from the force modulation can provide the needed distinction between
conductivity and topography that gets muddled when one images with the STM.
Figure 1.6 illustrates the usefulness of this separation. The image is of
bare, clean fluorine doped tin oxide surface measured in a non-conductive
solvent. The left image is the topography image, determined by scanning the tip
across the image in contact mode, maintaining a constant force. The tip was
coated with platinum and a bias was applied to the surface. The image on the
right is the current flowing through the tip/sample junction as the scanning occurs.
Both images were taken at the same time. It is clear from the images that certain
features in the topography are not conductive, while others show a high amount of
conductivity. In this way, nanoscale topography can be directly correlated with
Figure 1.6. Conducting AFM images of fluorine doped tin oxide. The left image shows the topography of the surface and the image on the right shows the simultaneously acquired nanoscale conductance map. The area is the same for both images.
the nanoscale electronic properties of the material. This is a function lacking in
the STM.
With conductive AFM probes and more complex instrumentation, more
sophisticated electronic AFM measurements can be performed. One
measurement mode that is gaining in popularity is Kelvin force microscopy
(KFM) which allows for nanoscale surface potential measurements. This
measurement is typically performed in tapping mode. In KFM an oscillating bias
is applied to the tip at a different frequency than the frequency at which the tip is
vibrated along with a servo controlled DC bias. If a surface potential exists, it
will cause the tip to oscillate at the amplitude of the AC bias due to changes in
electric fields produced between the tip and the sample. The amplitude of this
oscillation is typically very small and needs to be measured with a separate lock-
in amplifier. The amplitude is then negated by applying the appropriate DC bias 13
14
to the tip. This DC bias is then a quantitative measurement of the surface
potential.10
Instrumentation for AFM is typically a little more complex than for STM,
but is very similar. AFM cantilevers maintain constant force interactions by
pulling into or away from the surface through a mechanical connection to a piezo
element. Similarly, raster scanning can be achieved with the piezo in the AFM
scanner (attached to the cantilever) or by scanning the sample stage. AFM
scanning heads are equipped with laser and optical elements that allow for
reflection off of the backside of the cantilever onto a four quadrant photodiode
positioned somewhere above the tip. For imaging in amplitude modulation
dynamic force mode, a lock in amplifier is necessary in order to measure very
small changes in amplitude of cantilever oscillation at the vibrating frequency. A
preamplifier in the scanning head is necessary for conducting AFM
measurements, while a second lock-in amplifier is necessary for KFM
measurements to track changes in amplitude due to the oscillating tip bias.
Figure 1.7. SEM image of an AFM tip. The tip is positioned at the end of a cantilever. In this case the cantilever is triangular in shape. This tip is coated with gold for use in CAFM experiments.
AFM cantilevers (See Figure 1.7) are typically made out of silicon using
microfabrication techniques. Their size and shape determines their resonant
frequency and spring constants. Resonant frequencies can range from a couple of
kHz to several hundred kHz and spring constants range from fractions of N/m to
100 N/m. Low resonant frequency tips are useful for oscillating the tip in non-
contact mode in liquids, while high resonant frequency tips are standard for
intermittent contact mode. Low spring constants are useful when imaging in
contact mode, as the lower forces do less damage to both the tip and the sample,
while higher spring constant tips are usually used in intermittent contact mode.
The tip at the end of the cantilever ranges in radius of curvature from several
nanometers to several hundred nanometers depending upon fabrication method
and coatings and is chosen based upon potential application. Electrical coatings
15
16
for cantilevers can be made by sputtering or ion beam coating materials like gold
or platinum, while other conductive materials like boron doped diamond can be
created by CVD.11
1.1.3. Mercury drop electrodes
Although perhaps not as sophisticated a technique as scanning tunneling
or atomic force microscopy, mercury drop electrodes have proven to be efficient
tools in studying the electrical properties of organic molecules. Mercury is an
unusual element due to its very low melting temperature of 234.3 K12 making it a
liquid metal at room temperature. The liquid nature of the metal make it a
versatile electrode for use in various electrical measurements. Mercury drop
electrodes are commonly utilized in voltammetry experiments and are often
utilized to contact organic layers for conductance experiments. Its cohesive
nature and lack of adhesion to most materials make it well suited for forming very
flat reproducible contacts. Furthermore, thiolated molecules are able to quickly
form monolayers on a mercury surface due to interactions between mercury and
sulfur, allowing for a functional group which can make a direct chemical contact
to the electrode. This property is an important feature in high quality,
reproducible molecular electronics measurements, as will be discussed more
thoroughly in the following section.
1.2. Molecular electronics
Figure 1.8. Aviram and Ratner molecular diode. A) Chemical structure of the Aviram/Ratner diode. B) Energy diagram of the wired diode.
In 1974 Avi Aviram and Mark Ratner published a paper outlining the
design of an organic, single molecule rectifier.13 Much like a solid state diode,
this organic rectifier (see Figure 1.8) once properly connected would only be able
to pass current when biased in one particular direction. The rectifier was cleverly
designed using properties of organic chemistry. On the left side of the molecule
is tetracyanoquinodimethane (TCNQ) which is an electron acceptor, meaning that
it can accept another electron at relatively low potential into its lowest unoccupied
molecular orbital (LUMO). On the right side of the molecule is a
tetrathiofulvalene (TTF) group. This is a well known electron donor, giving up an
electron at low potential from its highest occupied molecular orbital (HOMO).
The two groups are connected by a single methylene bridge, so as to keep the
orientation of the molecule stable. Figure 1.8 (B) shows the energy level diagram,
17
18
once properly wired, for this molecular rectifier. The donor should be able to
transfer a molecule from its HOMO to the LUMO of the acceptor once the
voltage reaches a certain threshold in one particular direction, but not when the
bias is reversed. This means that the molecule should, theoretically, act in a
similar manner to a solid state molecular diode.
There are several reasons why the ability to make organic molecules work
in electronic devices is so intriguing. The versatility of organic chemistry, for
instance, could lead to the ability to design molecules to give the exact electrical
responses necessary for whatever device is being engineered. Simple
rearrangements of organic molecules could lead to fine tuning of electrical
devices at the most basic level. The potential of molecules as electronic
components however, goes far beyond their application as simple replacements
for current solid state electronic components. For instance, many molecules are
able to recognize other molecules through chemical bonding, or change behavior
upon changes in environment. This ability, coupled with the ability to electrically
transduce these signals can lead to the development of highly sensitive electrical
chemo-sensors. The device potential of molecular electronics is vast and specific
examples will be elaborated upon in later chapters.
1.2.1. Energy levels of electrons in organic molecules
The molecular diode that Aviram and Ratner created in their seminal
paper sparked much interest in the field of molecular electronics. What other
molecular devices could be created and what would their electronic behavior be
like? As the energy diagram in Figure 1.8 (B) indicates, the location of electron
energy levels in molecules is an important component to understanding how
molecules might conduct electricity. There are several theoretical approaches,
ranging from simple calculations to calculations that require intense computer
processing power that can give estimates on molecular energy levels.
Going back to quantum mechanics, the stationary state electronic energy
levels for a quantum system can be represented by
⟩⟨
⟩⟨=
ψψψψ H
E 1.2
Where ψ is the wavefunction and H is the stationary state Hamiltonian. One
method to determine electronic energy states is to use a “trial” wavefunction in
the equation for energy. The accuracy of the determined energy will be
dependent upon the trial function chosen. To illustrate this idea, the wavefunction
is written as a combination of basis functions
∑=i
ii fcψ 1.3
And this trial function can be plugged into the equation for the energy, giving
...
...
21211121
121221211121
+⟩⟨+⟩⟨
+⟩⟨+⟩⟨+⟩⟨=
ffccffc
fHfccfHfccfHfcE
1.4
19
Making ⟩⟨= jiij fHfH , ⟩⟨= jiij ffS and noting that H12=H21 for functions that
solve eigenvalue equations, equation 1.4 reduces to
...2...2
12211121
12211121
++++
=SccScHccHcE
1.5
With any set of given trial functions, the coefficients can be calculated by
minimizing this steady state energy. This is accomplished by taking the
derivative of the energy with respect to each consonant and setting that equal to
zero. This in turn leads to a series of equations that can be represented with the
following matrix determinant
022221212
12121111
=−−−−
OMM
K
K
SEHSEHSEHSEH
φφ
φφ
1.6
Depending upon the number and types of basis states involved, finding exact
solutions to this equation can be incredibly difficult.
A common method for determining energies of molecular orbitals is to use
a linear combination of atomic orbitals (LCAO) as a trial function for the
molecular orbital. One can determine the energy of the hydrogen molecular
orbital, for instance, by combining the 1s orbital wavefunctions of the electrons in
the two hydrogen atoms that combine to make the molecule. This combination
could then be plugged into the above equations to determine exact coefficients,
which eventually leads to the determination of two bonding orbitals with different
energies corresponding to the bonding and antibonding states of the molecule.
20
21
This method becomes significantly more difficult with atoms that have higher
valence.
For an atom such as carbon, molecular orbitals are produced by taking
linear combinations of the atomic orbitals for the electrons in the outermost
unfilled shell. For carbon this includes two 2s electrons, and two 2p electrons. A
molecule like CH4, however, makes four bonds that are equal in energy and
symmetry. Thus, the combination of atomic orbitals for CH4 will contain the
hydrogen 1s wavefunction, as well as 2s and 2p wavefunctions from the carbon
atom. This combination of both the s and p states of the carbon atom is referred
to as molecular hybridization and the resulting orbitals are referred to as sp3
molecular orbitals due to the use of all three p states (px, py, pz) and the s states in
the combination of atomic orbitals.
An unsaturated molecule like ethylene (C2H4) forms different bonds than a
molecule like ethane (C2H6). Ethane contains carbon that is sp3 hybridized like
the example of methane in the previous paragraph. In ethylene, however, each
carbon atom is bonded to two hydrogen atoms and contains a double bond to the
other carbon atom. The two bonds to the hydrogen atoms and one of the bonds to
the carbon are referred to as σ bonds, and are sp2 molecular orbitals, because each
orbital corresponds to mixing of the 2s carbon states and two of the 2p (px and py)
carbon states. The double bond is formed by the combination of the 2 pz states of
the two carbon atoms and the orbital is referred to as a π molecular orbital. The
two types of orbitals, σ and π, have different character. Figure 1.9 shows the
Figure 1.9. Orbital diagrams of sigma and pi bonding.
difference between the two types of bonding. In σ orbitals between carbon atoms,
the main sp2 orbital axis lie on the same line, and the σ bonding orbital is highly
symmetric, lying in the plane of the molecule. In the creation of the π orbital, the
p atomic orbitals have parallel axis, and overlap in a less symmetric way above
and below the atomic plane of the molecule. The lack of symmetry of the π bond
indicates that the carbon-carbon bond cannot be rotated, creating a more rigid
structure than is seen in ethane.
One simple means of examining electronic states in π orbitals in
unsaturated organic molecules, such as ethylene, is to make certain assumptions
about the quantities shown in equation 1.6. Those assumptions are that the
overlap integrals, Sij, are zero except for when i=j, Hii is the same constant (α) for
all carbon atoms, and Hij is zero for all but adjacent atoms for which it is equal to
22
β . These assumptions make up what is known as the Hückle model. For
ethylene, the determinant in equation 1.6 reduces to
0=−
−E
Eαββα
1.7
Which leads to βα ±=E , with β being negative for an attractive potential.
Plugging the value for energy back into the coefficient equations, gives the value
of the coefficients for the wavefunctions of the π orbitals. This process is
expanded for molecules with more carbon atoms, such as butadiene or benzene,
but is essentially the same.
Obviously, calculating electronic energy states accurately can get
complicated very quickly, even when certain terms are left out of the equation and
trial wavefunctions are utilized. For this reason, another method to calculate
energy states is often employed which does not rely on the use of wavefunctions.
This method is referred to as Density Functional Theory and was first postulated
as a method to determine energy levels and electron densities in materials by
Hohenberg and Kohn in 1964.14 In this theory an energy functional, E[n(r)],of the
electron density function n(r) is implemented to calculate energy states. The
value of this functional is determined at various states of the electron density until
a minimum is realized. This, then, gives a good estimation of electron density in
the ground state and ground state energy. Density functional theory is well
utilized in molecular electronics to determine energy states for determination of
electron transfer rates.15,16
23
24
1.2.2 Electron transfer
Electronic energy levels in organic molecules are only part of the story
behind molecular electronics. How electrons transfer from one part of a molecule
to another, or from molecule to molecule is of equal importance and questions
about the rate of electronic transfer are not simply understood by reading an
electronic energy diagram. For example, if there is a electron transfer reaction
that has zero free energy difference, electrons still do not randomly transfer from
one reactant to another as one would expect. The transfer must still be thermally
activated due to environmental factors. When an electron is on one reactant, the
solvent around that reactant is organized in a particular manner. In order to
transfer the electron to the other reactant, the solvent must reorganize onto the
other reactant; a process that requires energy not related to the electronic states of
the molecule. For this reason, thermodynamics play a large role in electron
transfer reactions.
Rudolph Marcus won the Nobel Prize in Chemistry in 1992 for his work
on the thermodynamics of electron charge transfer in molecular systems which
began in 1956 and is now known as Marcus theory.17 In his theory, energies of
the reactants and products of an electron transfer reaction were represented with a
single reaction coordinate, q, which represents all the parameters involved in the
electron transfer. Setting this value equal to zero when the electron is on the
product, the energy of the product is equal to
2
21 kqE = 1.8
This is quadratic because the leading term of an energy expansion about the
minimum is quadratic and higher order terms are ignored in this theory, assuming
that perturbations are small. The minimum energy of the reactants, then, occurs at
some other coordinate, qR so that the energy of the reactants is equal to
02)(21 GqqkE R Δ+−= 1.9
Where, is the difference in free energy between the products and reactants,
and k is the same as in equation
0GΔ
1.8 in order to make the following math easier,
noting that this is not typically the case for real systems.
This energy system is plotted in
Figure 1.10. Clearly, at some
point, q*, the energies of the
products and reactants are equal,
and for closely coupled, adiabatic
systems, once the reactant
thermally fluctuates to that reaction
coordinate, the electron will
transfer to the product. In the absence of thermal energy, however, the energy
needed to overcome solvation, the solvation energy, is λ. Solving for the point
that allows for thermally activated electron transfer, q*, gives
Figure 1.10. Marcus energy diagram.
25
)21(1 02* Gkq
kqq R
R
Δ+= 1.10
Leading to an effective activation energy of
2022
2** )21(
21
21 Gkq
kqkqG R
R
Δ+==Δ 1.11
Noting from Figure 1.10 that the squared sum in the energy can be related to the
solvation energy by
020
21 GkqG R Δ+=Δ+λ 1.12
The activation energy can be reduced to
λλ
4)( 20
* GG Δ+=Δ
1.13
Thus the rate of electron transfer is proportional to
)4
)(exp(20
TkGPBλ
λ Δ+−∝
1.14
This result clearly shows how, even if the free energy difference, , is zero,
the rate of electron transfer is thermally activated due to the solvation energy, λ.
0GΔ
1.2.3. Experimental methods for molecular electronics research
The Aviram Ratner diode was a rather revolutionary idea. However,
putting the idea into practice has proven difficult over the decades. Creating
stable, reliable contacts to molecules is a challenge to the field of molecular
electronics,18 as is maintaining the chemical stability of the molecules themselves.
The question of how to wire a single molecule to two electrodes for measurement
purposes has been answered by a variety of different techniques. It is important
26
27
to note, however, that no experimental technique is a perfect method for
examining the electronic properties of single or small ensembles of molecules.
Every technique has its own benefits and drawbacks, but the knowledge gained
from all of the experimental techniques gives a better view of actual electronic
processes in organic molecules, and they also point out the challenges in creating
future potential devices.
One early method for these types of measurements was to try to insert a
single or small ensemble of molecules between two electrodes that had a very
small gap. Needless to say, creating a gap that is large enough so that a molecule
can be inserted, yet small enough to keep anything larger than 1 or 2 single
molecules is tricky. One method of designing very small electrodes which has
been used with some success has been to drive a very higher current through a
thin piece of metal. This electric current is so strong that it actually moves the
atoms of the metal, and eventually creates a very small gap that stops the current.
This phenomena is known as electromigration. Once the gap is made, molecules
can be inserted into the gap and appropriate voltage can be applied. The
electromigration method of break junction creation has been used to study
transistor effects in fullerene,19 as well as Kondo effects (low temperature
conductance effects due to magnetic interactions) in divanadium.20 The
drawbacks of the method are inconsistent gap formations and the difficulty in
utilizing the created gap for multiple experiments.
Utilizing
scanning probe
techniques to
study molecular
electronics has
proven fruitful
over the last
decade. One
creative method
to wire molecules
to electrodes and
study molecular conductance has been to sandwich molecules between a
conductive substrate and a metal nanoparticle and to contact the nanoparticle with
a conducting AFM probe. Cui et al. demonstrated this method by creating a self
assembled monolayer (SAM) of octanethiol on a planar gold surface (see Figure
1.11). Octanedithiols were then inserted into this monothiol matrix. Gold
nanoparticles introduced to the surface selectively bonded to the isolated
octanedithiol molecules. Scanning these types of surfaces with a conducting
AFM showed images of the gold nanoparticles. The conducting AFM tip could
then be placed in contact with the metal nanoparticles and successfully create a
gold-molecule-gold junction through which to conduct current. The resulting IV
curves actually proved to be multiples of the lowest conductance IV curve,
Figure 1.11. Design and image of experiment to measure single molecule conductance using gold nanoparticles and conducting AFM.21 The left image shows the schematic of the experiment and the right image shows a typical AFM image with gold nanoparticles.
28
indicating that more conductive curves represented junctions with more molecules
in them. One noticeable feature in all the curves, however, was the lack of current
below a certain voltage in both forward and reverse bias directions. This was
attributed to coulomb blockade effect in the small nanoparticles. This coulomb
blockade effect was verified in later experiments utilizing larger nanoparticles that
have much lower charging energies.21 This method proved to be a reliable
method for measuring the conductances
of different molecules.23,24
Another widely utilized method
for determining the conductance of
organic molecules has been the STM
break junction method. Pioneered by
Xu and Tao25, this method has proven to
be widely robust and provides results
that are within an order of magnitude of
theoretical calculations of molecular
conductance.26 Using this method,
bifunctionalized molecules are either
attached to a metal surface or they are floating in solution above a metal surface.
An STM tip is brought into and out of contact with the surface repeatedly while
the current decay is measured (see Figure 1.12). When there is no molecule
trapped within the gap, the current decays exponentially with distance from the
Figure 1.12. Schematic of STM break junction experiment.25 The curves in the top left show steps from molecules in the electrode gap, while below it shows curves with no molecules. To the right are the resulting histograms from the two situations.
29
surface. If a molecule attaches to both the surface and the tip, however, the
current decay curves display very distinct plateaus or “steps.” These steps occur
because, once a molecule is stuck in the gap, the gold on the tip begins to stretch
as it pulls away from the surface. So, although the voltage on the piezo is still
pulling upwards, the distance of the gap remains unchanged, along with the
current. If all of the curves that have these plateaus are put together to form a
conductance histogram, there will be peaks where these steps occur. These peaks
are typically multiples of the conductance of the first peak, and correspond to the
conductance of 1, 2, 3… molecules stuck in the gap. This is a fairly easy and
direct way of measuring the conductance of a single molecule. Similarly, the
CAFM can be utilized to determine both the conductance of single molecules and
the force required to break the molecular junction simultaneously.27-29
Molecular electronics measurements can also be conducted using
the STM under potential control. The break
junction measurement, for instance, can be
used to understand how the conductance of
different molecules changes by applying
different potentials. This is done in electrolyte
solution with a reference and counter electrode
(see Figure 1.13). Visoly-Fisher et al. used this
method to study the conductance of
carotenoids, which are conjugated molecules
Figure 1.13. Schematic of STM under potential control.30
30
31
known to exist in photosynthetic complexes, under potential control.30
Carotenoids are thought to act as a means to discharge excess electrons that could
be potentially damaging in photosynthesis. The study showed that the carotenoids
become more conductive upon oxidation, which may be one of the first steps in
switching on this electron pathway. This method of potential control can also be
seen as creating an electrochemical “gate” and is used to study gating effects on
molecular systems.31,32
Another method of experimentally connecting molecules to two electrodes
has been to use a mercury drop electrode. In this method a self assembled
monolayer can be created on an electrode surface such as gold. A drop of
mercury can then be brought down onto this layer, and a voltage can be applied
between the mercury drop and the conductive surface. This process is relatively
more simple than the process of creating an electromigrated break junction, as
mentioned in the previous paragraph, since the junctions made are adjustable by
simply changing the height of the mercury drop. Furthermore, the mercury drop
can be used to study different positions on the same SAM surface. This method,
however, creates a larger electrode area, and, thus typically precludes single
molecule work. This method has been utilized to study conductance through
alkyphosphonate monolayers,33 as well as negative differential resitance effects in
molecules that have cyclic disulfide groups.34
32
2. DNA SEQUENCING BY RECOGNITION
Deoxyribonucleic acid (DNA) is a molecule that forms the basis for the
diversity in life seen on Earth. Every living cell contains this same chain-like
molecule, but the chain sequence differs from species to species and from
individual to individual. In humans, the sequence of nucleic acids (bases) in
DNA determines a multitude of traits, from relatively inconsequential
characteristics like eye color to serious potential conditions such as
predispositions to cancer. Despite the importance of DNA, there is yet to exist a
simple, cost effective method for determining the entire sequence of the human
genome. An advance in science that would bring the cost of DNA sequencing
down to $1000 per genome (a goal set by the National Human Genome Research
Institute which is a branch of the National Institutes of Health35) could lead to
great advances in personal diagnostics and personalized medicine. As the cost of
health care continues to increase, these advances may very well lead to more
effective, targeted treatment and prevention of disease, which could potentially
decrease the cost associated with the long term maintenance of many individuals
with diseases that, to date, can not be effectively cured.
Although there is currently a reliable method to sequence DNA, it is too
costly and too slow to be of practical use in determining the sequence of any
particular person’s genome. For this reason, many resources have been employed
to determine a better way to sequence DNA. One particularly interesting method
being researched is the idea of using the process of chemical recognition at the
33
single base level to determine genetic sequences. The structure of DNA will be
elaborated upon later in this chapter, however it is well known that DNA is
composed of complimentary strands of nucleotides made of four distinct bases.
Each base reliably hydrogen bonds with only one of the other bases to form the
highly recognizable double helix structure. The idea behind sequencing by
recognition is to utilize molecules that can reliably bind to the bases when the
DNA is in single strand form. A single molecular “reader” or several different
molecules could be employed that could interact with the different bases on a
single strand of DNA, and if this interaction could be transduced into a
recognizable signal that was base specific, it is possible that this could lead to a
high throughput method of DNA sequencing. Challenges to this method include
how to transduce the interaction between readers and bases into reliable signals
and how to construct a robust device that can easily read long single strands of
DNA by base recognition. To understand these challenges better, a basic
understanding of DNA structure and function is necessary.
2.1. Deoxyribonucleic acid (DNA)
Figure 2.1. Molecular structures of DNA components. The picture on the left shows how the DNA sugar rings are connected through a phosphate backbone. The middle images show the structure of the DNA bases, and the image on the right shows how these bases hydrogen bond with one another.
Figure 2.1 shows the chemical structures that make up DNA. DNA is
comprised of chains of 2'-deoxy-D-ribose (5 member sugar rings) connected
together through a phosphate backbone. Attached to each of the sugar rings is
one of four heterocyclic nitrogen compounds, commonly referred to as bases.
Two of these compounds, adenine and guanine, are substituted purines, while the
other two, thymine and cytosine, are substituted pyrimidines. Each base is able to
reliably hydrogen bond to one other base. Adenine hydrogen bonds to thymine
forming two hydrogen bonds, and guanine binds to cytosine forming three
hydrogen bonds (see Figure 2.1). In this way, complementary single strands of
34
35
DNA are able to bind to one another to form the well known double helix
structure.
The image on the left of Figure 2.1 shows how the sugar rings of DNA are
linked by the phosphate backbone. This image only shows the linking of three
sugar rings, but the sequence can go on to an infinite number of links. In DNA
nomenclature, the carbons on the sugar ring are labeled with a prime symbol,
distinguishing them from the carbons in the bases. Clearly, from the structure,
single stranded DNA has two different ends, the 5' end and the 3' end. In double
stranded DNA, the complementary DNA strands are bound together with
opposing ends. The 5' end of one strand of DNA matches with the 3' end of the
other strand. This is an important factor in the processes that lead to protein
assembly, as DNA is only read from the 5' end to the 3' end.
A base attached to 2'-deoxy-D-ribose (the sugar ring in DNA) is referred
to as a nucleoside. The nucleosides are then adenosine, guanosine, cytidine and
thymidine. A nucleoside that is bound to a phosphoric acid, either at the 5' or 3'
position is referred to as a nucleotide. Nucleotides are referred to by the
nucleoside, the position of the phosphate and the form of the phosphate.
Nucleosides can bind to a phosphoric acid in the monophosphate, diphosphate or
triphosphate form. For instance, adenosine bonded at the 5' position to a
triphosphate is referred to as adenosine 5'-triphosphate (ATP). Nucleoside
triphosphates are the building blocks of DNA synthesis, during which the other 2
phosphates cleave off. ATP, by itself, has important roles as energy carriers in
biological processes as well.
In the cell, DNA is
translated into another biopolymer
called ribonucleic acid (RNA).
DNA differs from RNA in two
ways. RNA has an OH group
attached to the 2' position of the
sugar. RNA also does not utilize
the base thymine. In RNA the thymine base is replaced with another substituted
pyrimidine, uracil. Figure 2.2 shows the difference between thymine and uracil.
For translation to occur, the double strand of DNA has to be unwound. Once the
DNA has been unwound, one of the strands, referred to as the template or
antisense strand, is read from the 3' end and complementary bases of RNA bind to
the DNA and are linked together. The resulting strand of RNA contains the exact
same information as is contained in the DNA since the strand is complementary.
This RNA strand is then able to move outside of the cell nucleus and play one of
several vital roles in protein synthesis.
Figure 2.2. Molecular structures of thymine and uracil. Thymine is utilized in DNA and uracil is used in RNA
36
Table 2.1. The genetic code. Each three letter base segment in RNA codes for a different amino acid, or codes for the end of transcription (stop codon).
DNA codes for 3 types of RNA. Messenger RNA (mRNA) codes for the
peptide sequence of a protein to be synthesized. Ribosomal RNA (rRNA) is a
structural component of ribosomes, which is where proteins are synthesized
within the cell. Finally, transfer RNA (tRNA) binds to and carries amino acids to
the ribosome as the mRNA is read. Within the ribosome, the mRNA is read, and
each three base segment either codes for one of the 20 amino acids found in
proteins or signals for the stop of protein synthesis. These three base segments
are referred to as codons. Table 2.1 shows the code for amino acids from RNA
bases. As the RNA is read in three base segments, tRNA brings in the associated
amino acid and a subunit of the ribosome binds the amino acids together as the
37
38
peptide chain grows. This process keeps on going until the a specific “stop”
codon is read.
Proteins serves many vital functions in living systems, and thus, the DNA
that codes for them is a precious structure in cells. For this reason, sequencing of
DNA has, and continues, to play an important role in biological sciences.
Furthermore, protein generation and regulation is imperative in maintaining
optimal health, and many diseases are linked to problems related to proteins. The
ability to sequence the human genome at a fraction of the current cost and time
could potentially lead to detection of future diseases and increased
implementation of individualized medical treatment.
2.2. Sequencing of DNA
Modern day DNA sequencing is almost always completed through a
process developed in the early 1970s by Frederick Sanger.36 The human genome
is very large with just over 3 billion base pairs,37 and so a high throughput method
is needed to sequence it in a reasonable period of time. One of the first steps in
the Sanger method is to chop the DNA into much smaller sequences. This is done
with an enzyme called a restriction endonuclease. There are several hundreds of
these enzymes and they each work by cutting DNA at very specific sequences.
After chopping, the DNA fragment to be sequenced is labeled with a small
piece of DNA that contains a tag that can be identified later (either 32P or a
fluorescent label). This strand is referred to as a primer and marks the point
where DNA synthesis will occur. All four 2'-deoxyribonucleoside triphosphates
39
(the building blocks of DNA synthesis), one of the four bases in 2',3'-
dideoxynucleoside triphosphate form, and DNA polymerase (the enzyme that
adds DNA nucleotides to the chain) are combined. Nucleotides will be added to
the primer strand until one of the 2',3'-dideoxynucleotides gets added with copies
of the labeled DNA fragment. Nucleotides are joined to the DNA chain by the 3'
OH group on the sugar ring at the end of the chain attacking the α-phosphorous of
a nucleoside triphosphate. Dideoxynucleotides do not have the hydroxide group
at the 3' position on the sugar, and thus cannot react with the next nucleotide to
further the chain. The point at which the 2',3'-dideoxynucleotide gets added is
stochastic. In a given reaction with multiple copies of the DNA fragment to be
sequenced with the primer, there will be an array of DNA fragments synthesized
with different lengths, all corresponding to where the 2',3'-dideoxynucleotide was
randomly added to the chain. This reaction is repeated using the same DNA
fragment to be sequenced with the other three 2',3'-dideoxynucleotides.
Once the sequence
has been reacted with all
four 2',3'-
dideoxynucleotides, the
synthesized fragments, all of
different lengths, are
separated by length. If 32P is
used to label the DNA, then
the fragments corresponding
to each of the four 2',3'-
dideoxynucleotides are
loaded into separate lanes of
a polyacrylamide gel. An electric field is applied and the fragments travel down
the gel distances that are relative to their lengths (masses). When the gel is
exposed to a photographic plate which is then exposed to radiation, the position of
each fragment shows up as a dark band. Since each lane contains fragments from
a different 2',3'-nucleotide added to the chain, each lane then represents all the
lengths along the original sequence where that particular nucleotide occurs. By
examining all four lanes together (see Figure 2.3.) one can determine by visual
inspection the sequence of the original DNA.
Figure 2.3. Autoradiograph gel for reading the sequence of DNA. Each track represents the lengths where each particular base occurs.36
As discussed previously, most modern day DNA sequencing uses a
variation on the basic Sanger method. For higher speed and more automated
40
41
reads, the DNA fragment to be sequenced is labeled with one of four fluorescent
dyes. Capillary electrophoresis is used to separate the different lengths of DNA
that were synthesized, and as these fragments exit the capillary, the fluorescence
signature is detected by laser excitation, and the sequence can be decoded very
quickly by software systems.38 Accuracy in these systems is very high and they
can now read as many as 1000 base pairs in a DNA fragment. However, at a
price of $0.50 per thousand bases sequenced,38 the price to sequence the human
genome would be approximately 1.6 million dollars, which is obviously far to
high to make genomic sequencing a modern day reality for individuals.
2.3. Sequencing by recognition
2.3.1. Benefits of sequencing by recognition electron tunneling
As explained above, Sanger sequencing works well for short reads of
DNA sequences, but sequencing the entire genome through this process, as
automated as it has become in recent years, is still a timely, expensive venture.
New approaches to DNA sequencing have been deemed necessary for high
throughput, cost effective sequencing of mammalian genomes. One method of
sequencing that has received a good deal of attention over the last few years has
been to pass single stranded DNA through a nanopore which can read each base
as it passes sequentially through. If the nanopore itself can read the DNA bases,
the costs of a nanopore sequencer may be drastically lower than traditional Sanger
sequencing.
42
Initially, when nanopore sequencers were being studied, it was thought
that the sequence of DNA could be read by examining ionic current through the
nanopore. Applying a voltage to the nanopore in the presence of an ionic liquid
creates an ionic current through the nanopore.39 In the presence of another
molecule, like DNA, this current reduces at times when the molecule is
translocating through the nanopore due to the fact that the bulkier molecule blocks
a fraction of the ions from getting through and thus reduces the current flow.40 It
was thought that if different DNA nucleotides blocked the flow of the ionic
current by different amounts, then the base sequence of DNA could be
immediately read from a current trace of the nanopore as the DNA translocates
through the pore.
At first, this process of ionic current readout appeared to be working for
RNA. RNA strands of a purine base had a different ionic current signature than
those of a pyrimidine base when passed through an α-hemolysin pore.41 More
interesting was the fact that RNA with two joining segments, one of a purine and
one of a pyrimidine, could be translocated through a nanopore and the point of
switching from one base to the other could be detected.41 The results, however,
were disappointing when using DNA. Different bases had very little difference in
current signature, and it was eventually determined that the difference noticed in
the RNA strands had more to do with different base stacking than with single
nucleotide current blockage.42 In fact, it was determined that the current signature
differences were actually due to geometric consideration of as many as 20
nucleotides in the nanopore at a time. This was, of course, a setback for nanopore
sequencing.
Sequencing by
recognition is a way to get
around the lack of
contrasting current signature
in a nanopore. In a
recognition nanopore device,
a molecule that could bind to
the nucleotides would be
attached to electrodes near
the nanopore. As
nucleotides exit the nanopore, they would interact with these “reader” molecules
and create a unique conductance signature. A scheme for such a system is shown
in Figure 2.4, where a reader molecule hydrogen bonds to a single nucleotide
while another molecule grabs the phosphate backbone. This could either be
accomplished with a single type of reader molecule that interacted with and gave
a different current signal for each DNA nucleotide with which it encountered, or
there could be separate reader molecules for each nucleotide and many copies of
the DNA sequencing strand could be translocating through many pores with a
different reader molecule at the end. Examining current signatures through all
pores would then give a base readout for the entire segment of DNA, but would
Figure 2.4. Schematic for a nanopore DNA sequencer utilizing molecular tunneling recognition.43
43
44
require consistent synchronization of the translocation events of all the DNA
strands. Of course, the future of sequencing by recognition is dependent upon the
ability to create reader molecules that can accurately identify single nucleotides
through a conductance signature. The following section and chapter explore how
different experiments can be utilized to determine conductance through molecules
hydrogen bonded to DNA bases.
2.3.2. STM measurements of sequence recognition
As the previous chapter explained, the ability to specifically recognize a
molecule and then transduce that recognition into an significant electrical signal is
a key benefit of many molecular electronic applications. One potential scheme to
sequence DNA by recognition elements is to have a binding molecule recognize
each base and produce a measurable and distinct electrical signal so that the
electrical signal can be used to read out the sequence of bases in DNA. The goal
is to produce a quick and efficient DNA sequencer. One of the biggest
challenges, however, is to determine the structure of a highly selective base
reading molecule, and, further, to determine how the base- reader complex
operates electronically in order to effectively calibrate a unique signal upon
recognition of a specific base.
The natural starting molecule to examine as a base reader is the
complementary base itself. For example, if one wanted a molecule that
specifically interacted with the base adenine and produced a recognizable
electronic signal, the first molecule to try would naturally be a molecule that was
45
structurally similar to thymine. Since adenine binds very strongly to thymine, it is
possible that measuring the current through an adenine-thymine bond may be
significantly higher than through the solvent or through a base mismatch, and this
current could be utilized as the electronic recognition signal.
There have been several reports regarding the utilization of the scanning
tunneling microscope to examine the conductance relationship between hydrogen
bonded base pairs. STM probes are very sharp, and thus, make excellent
electrodes for this type of single molecule work. Ohshiro and Umezawa utilized
the STM to study the electronic interactions between base pairs by imaging
substrates containing self assembled molecules of bases with a gold tip that was
functionalized with another base that could hydrogen bond with the bases on the
surface.44 While the bases could be imaged with an unmodified tip, contrast was
greatest when imaging a SAM of a single base with a tip modified with its
complementary base. Furthermore, single complementary bases were identified
in high contrast while imaging a surface comprised of a mixture of different
bases, meaning that the tip was able to recognize the complementary base and
distinguish it from non-complementary bases in the SAM matrix. In addition, the
sequence of a short segment of a peptide nucleic acid consisting of two different
bases could be identified utilizing a tip modified with one of the complementary
bases. Although the time scale involved with STM imaging makes it an unviable
means of sequencing DNA, the implications of these results are clear. The
transverse conductance of DNA associated with through-base tunneling is more
than the tunneling through an unbonded base, or through a non-complementary
molecule. This is evidence that DNA may be able to be sequenced by recognition
electron tunneling.
Ohshiro and Umezawa’s work with the STM showed the potential of
studying transverse electron tunneling through base pairs, and even showed how it
would be possible to identify the
sequence of a simple pattern of
peptide nucleic acid. Peptide nucleic
acid (PNA) differs from DNA,
however, in that it has no phosphate
backbone and includes amino groups
that are able to interact readily with a
gold substrate. Repeating this type of
work on DNA would prove slightly
more complicated. One method to
study the transverse base pair
tunneling interactions in DNA was to
utilize guanadinium ions (see Figure
2.5), which can hydrogen bond to phosphates as linkers between DNA and gold
substrates. The guanadinium monolayers were formed on the gold surface
through a thiol bond. DNA, both double and single stranded, can then be
introduced to the monolayer and forms densely packed layers on the surface,
Figure 2.5. Schematic of STM tunnel decay experiment on DNA oligomers.45 The guanadiunium on the gold surface hydrogen bonds to the phosphate backbone of the DNA, while the functionalized tip binds to the bases.
46
interacting strongly with the guanadinium.46 This system then provides good
electrical linkage between the DNA and the gold surface.
This DNA-guanadinium system can be utilized to study the interaction
between base pairs on DNA, as opposed to PNA in previous work. Adsorbing
single stranded DNA on the
guanadinium monolayer leaves the
bases open to study by the STM
probe. This system was studied by
He et al. by examining the electron
tunneling decay distances while
moving base modified and bare gold
STM tips into and out of close
contact with the DNA layer on the
surface (see Figure 2.6).45 For
surfaces that contained DNA of a
single base, it was found that the
tunneling current would extend
much farther (out to 2 nm) if the tip
was functionalized with a molecule that could base pair with the base on the DNA
than if the tip was bare or functionalized with an incompatible molecule (see
Figure 2.6). For example, for a layer of cytosine DNA, the current with a G
functionalized probe would extend out to 2 nm, but with a 2-aminoadenine (2-
Figure 2.6. STM tunneling decay curves for different base functionalized tips and a DNA-Guanadinium surface.45 The solid black lines show cures from bare tips on different DNA surfaces, and the lighter curves are decay curves from functionalized probes.
47
AA) probe it would only extend to 1 nm, and with a bare probe the current had
completely decayed by 0.5 nm. Similarly, for a surface of T-oligomer, the current
extended further with a 2-AA probe (2-AA can form 3 hydrogen bonds to
thymine), than with a G probe or a bare probe.
Integrating these curves gives the charge transferred during the retraction
process. Figure 2.7 (a) shows the charge transferred for a G probe on a
guanidium surface, T oligo (dT)45 surface and a (dC)45 surface, the charge
transferred for a 2-AA probe on the same surfaces, and the charge transferred for
a bare probe on the same surfaces. Clearly, the more charge is transferred in
systems of strong hydrogen bonding interactions (G-probe and (dC)45 and 2-AA
probe and (dT)45). From this data, it was determined that using a threshold of 2.2
Figure 2.7. Charge transferred during STM probe retraction. a) Histograms of charge transferred with different surfaces and different tip functionalization. b) Fraction of positive recognition events compared to C/T ratio in DNA oligomer for both G probes and 2-AA probes. 45
48
49
pC as a positive recognition event would yield a 50% recognition rate for G
reading C, and an 80% recognition rate for 2-AA reading T. To test the
specificity of these interactions, surfaces with oligos containing ratios of T and C
in blocks were utilized and the fraction of positive recognition events (Q>2.2 pC)
was plotted for each probe against the ratio of T and C in the oligo. The charge
transfer was remarkably good at determining the ratio of the two bases in the
oligos (see Figure 2.7 (b)) down to a block of target bases of 20. If the target
block was under 20 bases, reliable results were unattainable for reasons that will
be elaborated upon in the next chapter.
This tunneling decay method with the STM was further studied with
nucleosides (bases with sugar rings but no phosphate backbone) as monolayers on
gold surfaces connected through a thiol bond attached to each base.47 Probes
were functionalized with 2-aminoadenine, guanine, adenine or thiophenol.
Thiophenol does not hydrogen bond with DNA bases, but was utilized as a
control molecule. As with the DNA, it was noted that experiments that utilized
tips that hydrogen bonded with the target nucleoside on the surface had tunneling
curves that extended further than tips on bare targets or tips with molecules that
did not hydrogen bond to the target nucleoside. Furthermore, it was shown that
base combinations that made three hydrogen bonds extended even further than
base combinations that formed two hydrogen bonds.
These experiments were conducted by holding the tip at a setpoint current
and then retracting them at a given speed. Curiously, the setpoint current strongly
50
affected the shapes of these curves. For instance, a curve that began at a setpoint
of 2 nA was not simply the lower part of a curve that began at 3 nA. Taking a
polynomial fit from a curve at a higher setpoint and comparing the fit to the
curves at lower setpoints showed very clearly that there was a fundamental
difference between curves at different setpoints. Furthermore, the charge
transferred in these experiments was highly correlated with the initial setpoint.
These results, along with the fact that the decay distances were much longer than
typical electronic processes, and the fact that, upon approach, the tips had to be
brought much closer into the surface and had very steep current increases led to
the belief that the curves were representative of both electrical and mechanical
processes. This electromechanical system was studied further using the CAFM,
and results from those experiments will be shown in the next chapter.
3. ADHESION FORCE AND CONDUCTANCE MEASUREMENTS OF
DNA NUCLEOSIDES
Previous STM experiments on tunneling through DNA bases gave
preliminary evidence that the current signature of hydrogen bonding between
complimentary bases and nucleosides could be used as a recognition signal in a
nanopore molecular recognition sequencer. The data, however, left some
unanswered questions. The STM data on the DNA oligomers could only resolve
nucleotides that were in segments greater than 20 bases. Furthermore, the STM
withdraw curves on the
nucleoside surfaces varied
significantly as the current
setpoint changed and
extended much further out
than electronic processes
indicate. To demonstrate the
dependence of curve shape on
setpoint, Figure 3.1 shows the
STM data for a guanine
functionalized probe on a cytidine surface at several different setpoint. The curve
at 3 nA is fit with a polynomial whose function is used to fit the other curves at
different setpoints by simply shifting the function to the other current setpoint.
Figure 3.1. Initial STM retraction curve fits for a guanine functionalized probe retracting from a cytidine functionalized surface. The 3 nA curve is fit by a polynomial whose function is used to fit the other curves. Clearly, the function does not fit the other curves well.
51
52
The predicted curve simply does not fit any of the other curves with any degree of
precision, which clearly shows this change is curve shape.
Because of the discrepancy between what is expected for purely electronic
single molecule work and what was seen in the actual STM data, it was suggested
that these retraction curves represented more than electronic processes and there
were, in fact, geometric and mechanical factors influencing the measurements.
With the STM, however, it is difficult to separate mechanical and electrical data
in a single measurement since the tip is purely controlled by electronic tunneling.
It is easier to understand how mechanical factors influence an electrical
measurement if the measuring device has the ability to separately monitor
mechanical effects. For this reason, the conducting AFM was utilized to
understand the electromechanical properties of the base-nucleoside system.
Chapter 1 explained some of the benefits of utilizing the conducting
AFM. Specifically, in this case, the same measurement that was conducted
utilizing the STM can be performed on the AFM. However, in the case of the
AFM, the force on the flexible cantilever can be recorded simultaneously with the
electrical current. By comparing mechanical force with different attributes of the
electrical curves, one can make inferences about what might actually be occurring
as the nucleobase on the tip binds with the nucleoside on the surface and the
complex is pulled apart. This information may be of use in designing future DNA
sequencers.
3.1. CAFM measurements
53
3.1.1. Materials and methods
A picture of the
experiment is shown in Figure
3.2. Gold on Mica substrates
were received from Agilent
Technologies (Chandler, AZ).
They were annealed with a
hydrogen flame to form clean
<111> surfaces directly before
assembly of nucleoside monolayers. Thymidine self assembled monolayers were
created by first making a 1mM solution of 5'-S-acetyl-5'-thiothymidine in
methanol, adding 10-20 drops of pyrollidine and allowing to sit for 1 hour to first
remove the acetyl group from the molecule. After the hour, a freshly annealed
gold substrate was left in the solution for another hour. The thymidine bonded as
a monolayer to the gold surface through a thiol group at the 5' position of the
sugar ring. After an hour, the substrates were rinsed with methanol to remove
molecules that were not firmly bonded through the thiol linkage and were then
dried with nitrogen.
Figure 3.2. Picture of STM/AFM experiment. A gold probe with a nucleobase is brought into contact with a nucleoside surface and then retracted.47
Figure 3.3. SEM image of AFM tip radius after metallic coating.
AFM tips used for CAFM experiments were from MikroMasch, (Tallinn,
Estonia) and had reported spring constants of 0.35 N/m with triangular
cantilevers. They were coated with an ion beam coater from Gatan (Pleasanton,
CA) with chromium and gold in the following layers (25 Cr, 50 Au, 25 Cr, 50 Au,
25 Cr, 100 Au, 25 Cr, 300 Au). This layered structure of chromium and gold
produces a more robust electrical coating on the AFM tips that is much less likely
to peel off during measurements. The radius of curvature of a typical probe after
coating was 100-200 nm and was measured by scanning electron microscopy
(Figure 3.3). Probes were immediately transferred from the coater where they
were held in vacuum to solutions of either 1 mM 8-mercaptoadenine or 2-amino-
8-mercaptoadenine in dimethylformamide or 2 mM thiophenol in methanol for 2-
12 hours. All molecules for tip functionalization were received from Sigma
54
55
Aldrich (St. Louis, Mo). Before use, the tips were rinsed in DMF and then
methanol and dried in air or Argon.
All experiments were performed in 1,2,4-Trichlorobenzene solvent, which
is a slightly polar but non-conductive solvent, in a Teflon liquid cell that had been
cleaned with piranha (3:1 sulfuric acid: hydrogen peroxide). Functionalized tips
were approached to the SAM surface using the Agilent PicoScan software on an
Agilent PicoAFM microscope. After successful approach, custom labview
software was utilized to maintain the tip at a current setpoint chosen by the user
with a software PID controller. Upon withdraw, the servo was broken and the tip
was retracted at a speed of 2,000 nm/s. Both current and cantilever deflection
(from the AFM photodetector) were recorded through a breakout box on a
Yokogawa DL750 Scopecorder oscilloscope, (Houston, TX) saved on the hard
drive and transferred to computer files later.
After experiments, the spring constants of all tips were measured using the
thermal noise method48 on an Agilent PicoPlus microscope and a National
Instruments PXI 6120 data acquisition card. Although tips are ordered at a given
spring constant, they can vary by an order of magnitude from tip to tip within the
same batch. Furthermore, the metallic coating of the AFM tip changes the spring
constant (see Figure 3.4 (d)). For force curves to be accurate, therefore, the spring
constant of every cantilever has to be measured. This is accomplished by
examining the energy thermally dissipated in the cantilever while it is away from
the surface. The deflection of the cantilever is measured with the AFM
photodiode for a given period of time at very high bandwidth (typically at least 10
times the resonant frequency of the cantilever). A power spectral density of the
deflection trace is then measured and the area under the resonant frequency peak
is determined. The sensitivity of the cantilever in V/nm is also determined by
examining the contact region of a deflection approach curve on a hard substrate.
These two values combined can give the spring constant of the cantilever through
the following equation49
2
2 costan21
tan231
8174.0
⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢
⎣
⎡
⎟⎠⎞
⎜⎝⎛−
⎟⎠⎞
⎜⎝⎛−
= φφ
φ
LDLD
PsTK
k B
3.1
56
Figure 3.4. Procedure for obtaining cantilver spring constant. Cantilever deflection noise is recorded (a), followed by conversion to a power spectral density (b). Finally the resonant peak is fit with a lorentzian (c) in order to compute the area under the curve. This value, combined with the cantilever sensitivity is used in equation 3.1 to determine the spring constant. (d) shows the average spring constant values for several cantilevers before and after chromium and gold coating. Where s is the sensitivity, P is the area under the power spectral density resonant
peak and D, L and φ are geometric measurements of the cantilever typically
obtained from the manufacturer. A custom labview program was created to better
automate the thermal noise measurement and obtain spring constants for all
cantilevers used in the experiments. Figure 3.4 shows the steps necessary to
compute this spring constant.
Data was analyzed using custom labview analysis software. Deflection
curves were translated into force curves in nN by using the measured spring
constant and the sensitivity of the tip in V/nm as calculated from each individual
force curve in the contact region. Figure 3.5 shows what a typical force retraction
curve looks like, along with the simultaneously obtained current curve. The linear
57
region in the force curve occurs when the tip is in contact with the surface. As the
tip moves away from the surface, recorded force decreases linearly due to the
spring nature of the cantilever. The force becomes negative due to adhesion of
the tip with the surface and finally breaks away to zero force by the end of the
retraction. The minimum force value was calculated by the labview program and
is referred to as the adhesion force. Similarly, for the current curves, the value of
the current before withdraw and at F=0 are recorded by the labview software and
are correlated with adhesion force.
Figure 3.5. Typical force and current curve. The blue curve is the force while the red curve is the current. The two were obtained during the same withdraw and were recorded simultaneously. Fad is the adhesion force felt by the tip during retraction.47
3.1.2. Relationship between force and setpoint current
The force of adhesion on a cantilever during retraction from a surface is
due to a number of different factors. On a bare surface with a bare tip in a liquid
58
59
environment, the adhesion force is usually very small. In air, adhesion is
significant due to a water layer that forms on the surface. In the case of the
nucleosides, adhesion can be due to non-specific interactions between the
molecules on the tip and the molecules on the surface and due to the binding
between complimentary base pairs. Measurements for each functionalized tip
were taken at several different setpoints, and for each setpoint, the average
adhesion force and current at retraction were recorded. The current at the point of
retraction slightly differs from the setpoint current since the PID controller is not
very accurate and the current actually fluctuates considerably around the setpoint
value. The retraction current and adhesion forces were plotted against one
another. This plot is shown in Figure 3.6.
Figure 3.6. Plot of adhesion force versus the current at probe retraction. The surface was a monolayer of thymidine while the tips were functionalized with 2-aminoadenine, adenine or thiophenol.47
From Figure 3.6 it is clear that the adhesion force increases faster while
using a 2-AA probe (which forms three hydrogen bonds with thymine) than with
the A probe (which forms two hydrogen bonds with thymine). Both tips have
much larger forces than seen with the thiophenol probe, which does not actually
bind to the thymine. Since the thiophenol is similar in structure to the other
molecules it is a good control as it will have similar non-specific adhesion to the
SAM, without the bonding. The relationships are well-fitted with a natural
logarithm, as shown in the figure.
60
The fit of the AFM data to a logarithm coincides with a multiple-molecule
interaction model. The number of molecules in the interaction is strongly related
to the contact area of the junction. From Hertz theory, the contact radius is
related to the displacement of the probe by50
Ra 2
=δ 3.2
Where δ is the indentation of the probe, a is the contact radius and R is the radius
of the probe. The tunneling current increases with the probe indentation, which
leads to the following relationship
)1( −= δbeAI 3.3
Where A and b are both constants and the current is set to zero when there is no
probe indentation. The number of molecules being probed, N, in any given
retraction is proportional to the contact area at the setpoint, which means that it is
also proportional to the current at the setpoint.
)1ln( += SPBICN 3.4
Where, again C and B are constants. The adhesion force is then proportional to
the number of hydrogen bonds, in addition to the nonspecific adhesion force.
This leads to the following relationship for adhesion force and setpoint current:
NSSPHBHBadhesion FBISNF ++∝ )1ln( 3.5
In this case, NHB is the number of hydrogen bonds per base pair interaction (2 for
adenine-thymidine and 3 for 2-aminoadenine-thymidine), SHB is the strength of
each hydrogen bond and FNS is the nonspecific adhesion which can be calculated
61
from the thiophenol control. This is the function utilized to fit the data in Figure
3.6, assuming that all hydrogen bonds are the same strength (5-7 kcal/mol).51 In
fact, the coefficients in front of the logarithm were fit for both A-T and AA-T.
The ratio between the two coefficients was 1.92 (+0.7,-0.5) which is very close to
the 1.5 expected from the ratio of hydrogen bonds (3:2) in each interaction.
This relationship,
discovered from the
conducting AFM data, has
led to the elastic interaction
model for the STM
retraction curves. A
schematic of the model is
shown in Figure 3.7 and is
valid for both the AFM
experiment and the STM
experiment. K2 represents the stiffness of the probe-molecule bond, while K1
represents the stiffness of a single hydrogen bond. X1 is the distance of the tunnel
gap, while XP represents the external pull of the probe measured by voltage on the
piezo. Changes in XP actually results in smaller changes in X1 due to the
mechanical interactions of the junction. Since XP=X1+X2, X1 can be related to XP
by,
Figure 3.7. Schematic of the elastic-distortion model.47
62
PG
XKK
KX+
=2
21
3.6
Where KG
is the spring
constant of the
molecular gap,
which is defined
as the product of
the number of
formed hydrogen
bonds and the
stiffness of those
bonds, K1. This
clearly explains
why the STM tunneling curves decay so far. One would expect, then, that probes
that are less stiff have even longer decays since K2 would be dominated by the
low probe stiffness. This is seen in comparing the AFM and STM withdraw
curves. The AFM probe at k=0.35 N/m is much more flexible than the STM
probe, and so the current retraction of the A-T systems extends further out to 4 nm
as compared with the STM curve which extends out to 2 nm. This can be further
examined with the conducting AFM by changing the spring constants of the
probe. Figure 3.8 shows how going to a higher spring constant probe decreases
Figure 3.8. Decay curves for AFM cantilevers of different spring constants. The stiffer cantilever begins to approach the STM data47
63
the extension of the current. A 2 N/m AFM probe begins to emulate the extension
of the STM probe.
Since KG represents the stiffness of the molecular junctions, which is
directly proportional to the number of molecules being probed times the product
of the number of hydrogen bonds per molecular interaction and K1, the stiffness
of a single hydrogen bond, then equation 3.6 can be rewritten as
PSPHB
XBIKKCN
X)1ln(/1
1
211 ++= 3.7
The parameters needed to fit
this model are CK1/K2 and B.
These values can be readily
obtained by taken a subset of
the data at a particular decay
value. For instance, the
distance where the current
falls by a factor of 2, XP(0.5),
can be obtained from the data
at a number of different
setpoints, and CK1/K2, B and
X1(0.5) can be fit accordingly (see Figure 3.9).
Figure 3.9. Elastic-distortion model coefficient fitting.47
Equation 3.7 can be utilized to fit the STM current decay curves. First, an
arbitrary function can be assigned such that
64
))1ln(/1
()()(21
11 ++==
SPHB
PSPSP BIKKCN
XfIXfIXI ββ3.8
Where β is the inverse decay length of the curves in solvent and f, again, is an
arbitrary function equal to 1 at X1=0 and 0 at X1=∞. If the function can be
calculated given a set of XP at a particular setpoint, then a simple change in
coordinates will allow the function to be utilized for other setpoints by the
following relation
)()1ln(/1)1ln(/1
)( 2221
1211 SPP
SPHB
SPHBSPP IX
BIKKCNBIKKCN
IX++++
=3.9
Figure 3.10. STM withdraw curves for G-C and A-T interactions. Dashed lines represent a polynomial fit for the 3 nA setpoint curve, while the solid lines represent the same polynomial fit, but using the elastic-distortion model to transform the displacement coordinate.47
Equation 3.9 can now transition coordinates to fit the STM decay curves properly.
If a function can be found that can fit one of the curves starting at one setpoint,
then the other curves can be fit with that same function by making the coordinate
conversion in equation 3.9. This was accomplished for both the STM G-C
65
interaction and the A-T interaction shown in Figure 3.10. Both plots show several
curves beginning at different setpoints. The dashed line is representative of a
polynomial fit to the 3 nA setpoint curve used to fit all of the curves without
coordinate change. The solid green line is the same polynomial fit with
coordinate conversion from equation 3.9. Clearly, the solid green line is a much
better fit to the data, giving evidence that this elastic-interaction model is a good
representation of what is actually occurring in the measurement.
This elastic-interaction model was also used to fit STM data related to the
charge transferred during tip retraction. The charge is found by finding the area
under the retraction curve, which is equivalent to integrating the current as a
function of time. With the arbitrary function in equation 3.8, this works out to
∫ ∫∞ ∞
++==
0 0 21
))1ln(/1
()( dtBIKKCN
vtfIdttIQSPHB
SPβ 3.10
Where υ is the retraction speed of the probe. This, after substitution, works out as
))1ln(1(2
1 ++= SPHBSP BIN
KCK
vI
Qβ
3.11
66
This formula was used to fit
data on charge transferred
versus setpoint current, and
is shown in Figure 3.11. It
is suspected that the small
variations from this theory
are due to the slight
difference between
hydrogen bond strengths
between the different base pair combinations.
Figure 3.11. Elastic-distortion model fits to the STM charge transferred data.47
3.1.3. Calculations of molecular conductance
Figure 3.5 shows actual force and current curves obtained simultaneously
in an A-T CAFM interaction. The point at F=0 in the retraction is shown by an
arrow. This is, ideally, where the assembly has zero force applied, and would be
the best point to measure a conductance value for the molecular junctions. One
has to know, though, how many molecules are being probed in a given pull. This
is easily read from the adhesion force value, minus the non-specific adhesion. In
general, the AFM tip probed 2-10 molecules during a given retraction. Dividing
this value by the force to break the molecular bonds, which is 600 pN for a 2AA-
T pair and 400 pN for an A-T pair,52 gives a good estimate for the number of
molecules in any given pull. Based upon the number of molecules probed as
calculated from the adhesion force and the current at F=0, it was found that the
67
68
average conductance of a single molecular nucleoside-base pair was between 100
and 300 pS. This was an order of magnitude lower than theoretical calculations
(See Table 3.1) for nucleoside-base pairs, which is a standard discrepancy
between molecular electronics theory and experiment.26
3.2. Conclusions
In this experiment, the conducting atomic force microscope was an
important component in understanding the behavior of the mechanical and
electrical interactions between nucleoside-base pair bonding and probe molecule
interactions. The conclusion after AFM experiments was that the STM retraction
curves do not represent single molecule measurements. Going back to previous
STM experiments performed on DNA oligomers, it was noted that the base
functionalized STM tip did not have the ability to distinguish blocks of
nucleotides that were less then twenty in length. New understanding of the
experiment explains the reasoning behind this lack of discrimination. Based upon
this model, the STM tip was probing too many molecules to be able to accurately
distinguish such a small block of bases.
The elastic-distortion model obtained after AFM results also explains why
STM retraction curves have shapes that are dependent upon initial current
setpoint, and why charge transfer depends strongly on this setpoint. Furthermore,
the CAFM gave strong evidence that the conductance of these molecular
junctions is very low, leading to the belief that the initial current setpoint is
actually achieved through a through-space tunneling mechanism and not through
Table 3.1. Calculated conductances of base-base molecular junctions and base-nucleoside junctions.47
the bonds. The current setpoint does, however, affect the number of bonded
molecules in each retraction, which leads to the extended decay length and
increased charge transfer.
The final implications of these experiments relates directly into the
creation of a device that can eventually sequence DNA by recognition electron
tunneling. Clearly, from the AFM data, conductance in base-nucleoside junctions
is very low. This is supported by theoretical calculations which show that base-
nucleoside junctions have a conductance value that is an order of magnitude lower
than similar base-base junctions. Furthermore, calculations show that the
conductance differences between different nucleoside-base combinations is quite
small compared to differences seen in base-base interactions (Table 3.1). The
sugar rings associated with the nucleosides clearly dominates the conductivity of
the interaction. Adding a phosphate makes the conductance even lower. These
results suggest that a single base may not be able to discriminate nucleotides
coming out of a nanopore. Other readers or different schemes may be necessary
in order to utilize molecular recognition tunneling in effective DNA sequencers.
69
70
4. ORGANIC PHOTOVOLTAICS
Another application of molecular electronics that is drawing more interest
is within the field of organic photovoltaics. Traditional sources of energy in use
today are mostly derived from non-renewable fossil fuel sources. These sources
are in limited quantity and are often in politically unstable areas of the world.
Furthermore, more evidence indicates that the burning of fossil fuels is
contributing to global climate change. For these reasons, there is great interest in
renewable and cleaner energy sources. One obvious and highly abundant source
of renewable energy is solar energy, and there is much interest in harnessing the
power of the sun with photovoltaic technology. The sun is a near-infinite source
of energy, that, if converted properly, can provide society with the energy needed
to move forward in 21st century. The current problems with solar energy,
however, lie in lack of efficiency and cost.
Most photovoltaics currently utilize inorganic semiconductors like silicon
to both absorb light and transport photogenerated current. With the latest
technology in semiconductor processing and manufacturing, these devices can be
made relatively efficient, but at great cost. For the most part, semiconductor solar
cells available to the average consumer produce energy at a cost that is much
higher than fossil fuel sources. For example, the United States Department of
Energy (DOE) recently published that the levelised cost of energy for rooftop
photovoltaics in 2008 was between $0.20 and $0.80 per kWh.53 This is quite
expensive compared to the average retail price of electricity in 2008, which was
71
$0.11 for residential consumers.54 For this reason, research in organic molecules
as photovoltaic materials has grown over the past few years. Organic molecules
are utilized everyday for light energy conversion in plants, so it has been proposed
that they can be used efficiently in organic photovoltaic devices, producing less
costly solar cells. Although efficiency has remained an issue in organic
photovoltaic research, the prospect of a less costly device has continued to spur
more investigation in the field.
4.1. Photovoltaic energy conversion
There are two main processes occurring in the typical photovoltaic device.
The device must first absorb light and excite electrons into higher energy states,
and then those electrons must be transferred into electrodes. If wired in short
circuit form, this produces a photogenerated current upon illumination. If the cell
operates in open circuit design, this generates a photovoltage. The combination of
short circuit photocurrent and open circuit photovoltage gives the maximum
potential power that a solar cell can generate. Both the process of absorption and
transport are critical to a solar cell’s potential as a viable energy conversion
device. If the cell does not efficiently absorb at high intensity solar wavelengths,
then the cell has little utility. Furthermore, if those excited electrons are not able
to be transported, they eventually recombine and the absorbed energy is lost.
Balancing good absorption with good transport is a critical challenge in
photovoltaic design.
4.1.1. Semiconductor photovoltaics
Figure 4.1. Schematic of a P-N junction photovoltaic device. A similar device can be made utilizing a P-I-N architecture.
Figure 4.1 shows a schematic of a traditional silicon photovoltaic cell.
The cell operates as a P-N junction, with p-type semiconductor in contact with n-
type. When the two types of materials are in close contact, holes from the p-type
combine with electrons from the n-type and create a volume that is charge neutral
and is referred to as the depletion region. An electric field exists in the region
from the p-type to the n-type due to the difference in charge that is now separated
by the length of the depletion region. When light is absorbed in this depletion
region the electron and hole pair (the exciton) that is created is driven by the
electric field to further separate into the respective types of semiconductor. In
open-circuit configuration this creates a photovoltage, while in short circuit
configuration, this creates a photocurrent. The equivalent circuit for this device is
72
shown in Figure 4.2 including both the shunt resistance and series resistance of
the device.
Semiconductor
photovoltaic cells are
currently on the
commercial market,
available to both
residential and industrial
consumers. Many power
companies even allow
photovoltaics to be tied to the existing grid and buy any excess electricity that the
consumer does not utilize. Although semiconductor photovoltaics are making
headway in becoming larger-scale energy production sources, the drawbacks of
their use currently stops many consumers from purchasing. The efficiency of a
cell depends upon the quality and processing of the semiconductor, which is often
proportional to the cost of the cell. This expense, compared to the low upfront
cost of more conventional energy sources, can make consumers reluctant to
purchase photovoltaics.
Figure 4.2. Equivalent circuit for a photovoltaic diode device.
4.2. Benefits of organic photovoltaics
There are several different methods being utilized to both improve the
efficiency of photovoltaic devices and bring their costs down. One promising
method of producing low-cost photovoltaics is to utilize organic molecules as
73
74
light absorbers in a photovoltaic junction. At a very basic level, an organic solar
cell consists of a layer of organic light-absorbing material sandwiched between
one transparent electrode and another back electrode. As the organic absorbs
light and creates excitons, these exitons would migrate to the electrodes and
produce photovoltage or photocurrent. This type of scheme has several benefits
and holds promise for future commercial devices.
The main benefit foreseen with utilizing organic materials for photovoltaic
devices is in cost. If organic materials could be spin coated or self organized from
solution onto electrodes, the cost of processing compared to the cost of
semiconductor processing might be dramatically lower. The low cost of
processing, alone, may make organic photovoltaics good competitors for
semiconductor solar cells, even at lower efficiencies. Since much organic
photovoltaic (OPV) research involves the processing of organics in low cost
ways, this is clearly a goal for the OPV field.
Low cost processing is not the only potential benefit to organic
photovoltaic research. Organic molecules are versatile materials, and thus can be
changed and modified to suit different needs. Chemical variations in molecules
change how they absorb light and this can be utilized to tune the absorption of the
photovoltaic to the solar spectra in order to more efficiently capture available
radiation. Bulk material flexibility is also seen as a potential benefit of OPVs.
Unlike most inorganic semiconductor solar cells, organic thin films can be quite
flexible and are usually easily prepared on a variety of substrates. This leads to
the idea that organic films can be made on flexible substrates, which may
eventually lead to flexible photovoltaics. Since tradition semiconductor
photovoltaics are typically rather rigid, they are limited in where they can be
placed and how they can be transported. A flexible photovoltaic has many
commercial applications and is a key potential benefit to organic solar cells.
4.3. Challenges to commercial organic photovoltaics
Figure 4.3. Best research cell efficiencies by type of solar cell and year.55
75
Although the benefits of organic photovoltaics certainly make them
promising candidates for commercials solar cells, there are certain challenges that
have and continue to keep them from large scale commercial markets. Most
challenges in the design and fabrication of these photovoltaics is centered around
the low efficiencies of such devices. Figure 4.3 shows the efficiencies of different
types of research solar cells versus the year that the efficiency was determined.
Clearly, the graph shows that organic photovoltaics are among the lowest
efficiency devices. Figure 4.4 gives a schematic that explains the different
processes that occur in a typical organic device and is a good representation to
explain why efficiencies in organic photovoltaics are traditionally very low.
Every photon that gets absorbed by the film leaves an electron in a higher energy
excited state and a hole. In order to produce photocurrent or photovoltage, this
electron and hole pair need to travel to their respective electrodes, which are
determined by the electronic work functions. The efficiency of the device can be
determined by the rates
of carrier transport minus
the recombination rate.
Unfortunately, in
most organic devices the
rate of absorption is
small (organic molecules
typically have absorption
cross sections on the
order of Å2), while the
rate of recombination is
relatively high. Combined with low mobilities (as low as 10-4 cm2/(V*s) for
electron and hole mobilities for some polymer-fullerene blends discussed later in
the chapter)56 which limit the rate of particle capture by the electrodes, and the
Figure 4.4. Energy diagram of organic photovoltaic device.
76
77
efficiencies of these devices turn out to be low. Exacerbating the problem is the
fact that most changes that can improve one part of the efficiency equation tend to
negatively affect the other parts. For example, increasing the thickness of the
organic layer does increase the amount of light being absorbed, but a fewer
percentage of the excitons generated get to the electrodes. This is because
excitons must diffuse to the electrodes. The diffusion length before
recombination of excitons in typical organic film is on the order of 5-20 nm.57-60
So, a film that is any thicker than 20 nm does not collect all the generated
excitons due to charge recombination. Decreasing the thickness of the film
increases how many excitons make it to the electrodes, but obviously decreases
the amount of light absorbed. So beneficially changing this rate equation in
significant ways requires more sophisticated methods.
4.4. Organic photovoltaic architecture
As previously mentioned, organic photovoltaics suffer from low
efficiencies that are a result of competing factors in the rate equation that
determines efficiency. There have been numerous methods to increase the
efficiency that go beyond simply changing the thickness of the organic layers,
which, as mentioned, can only affect efficiencies by a very limited amount. For
this reason, several more sophisticated methods of producing more efficient
organic photovoltaics have been utilized with some success, although efficiencies,
and practicalities in some cases, have meant that these breakthroughs have not
been incorporated into commercial solar cells to date.
78
4.4.1. The Gratzel electrochemical cell
One exceptional improvement on the organic photovoltaic came in 1991
with the creation by O’Regan and Gratzel of an organic electrochemical
photocell.61 A schematic of a typical Gratzel cell is shown in Figure 4.5. The
original Gratzel cell utilized nanoporous TiO2 on top of sheets of conducting
glass. The TiO2 was then exposed to an organic dye which adsorbed onto the
surface. The resulting TiO2 was able to absorb up to 46% of incident solar light.
The dye-sensitized TiO2 was then put into contact with an electrolyte solution
such as iodide/triodide which was in contact with another transparent electrode,
like ITO. In this solar cell configuration, when the dye absorbs light, it injects the
excited electrons into the TiO2, leaving the dye positively charged. The
electrolyte can then quickly donate an electron to the dye leaving it neutral, and
leaving the electron permanently injected into the TiO2. The ITO then donates an
electron to the electrolyte ion in the solution. By connecting the TiO2 through a
load to the ITO in contact with the electrolyte, the circuit can be completed and
power generated.
This solar cell scheme
gets around some of the typical
problems with many OPV
devices. There is no need, for
instance, for the exciton to
diffuse within the organic, since
the electron can immediately
inject into the TiO2 and the hole
can be filled by the dye being in
constant contact with the liquid
electrolyte. Rates of electron
and hole transfer increase
without changing the amount of absorbed light. In addition, the nanoporous TiO2
provides a very high surface area electrode, allowing for more organic dye to
absorb onto the surface. This, then, increases the absorption of the device,
without creating a thick, non-conducting organic film. Originally these cells had
efficiencies from 7.1%-7.9%,61 which, amazingly, is still higher than thin film
organic photovoltaic devices. Current research in Gratzel-type cells has led to
photovoltaics achieving up to 10.6% efficiency.62 Gratzel cells, unfortunately, do
suffer from difficulties in practical application. They currently require a liquid
electrolyte in order to operate. This necessity has made it impractical to use in
any significant way as a photovoltaic device. A breakthrough in electrolyte
Figure 4.5. Gratzel electrochemical photocell
79
80
technology to create an efficient semi-solid electrolyte may lead the way into
bringing Gratzel cells into commercial use, and some research in this area appears
somewhat promising.63
4.4.2. Heterojunction organic photovoltaics
The Gratzel cell gets around some of the problems with recombination by
providing a simple pathway for excited state electrons to be donated directly into
a semiconductor. Getting the electron away from the excited state light absorber
is an essential component in efficient OPV devices. Another method of
accomplishing this is to utilize molecular donors and acceptors within the same
organic device. By having them close to each other in the same device, light
absorbers can quickly donate electrons into the acceptors, slowing the rate of
recombination and increasing the chance of extracting charges more efficiently.
This donor-acceptor scheme has had particular success when utilizing a
polymer like regioregular poly(3-hexylthiophene) (P3HT) or poly(2-methoxy-5-
(2'-ethyl-hexyloxy)-1,4-phenylene vinylene) (MEH-PPV) as the light
absorber/electron donor, and the fullerene derivative, phenyl-C61-butyric acid
methyl ester (PCBM)64 as the electron acceptor. The absorber and the fullerene
are typically blended together and spin coated onto an ITO substrate pre-coated
with Poly(3,4-ethylenedioxythiophene) :poly(styrenesulfonate) (PEDOT-PSS).
Efficiencies in the 3-6% range have been achieved by several different groups
utilizing this method.65-67 Since charge transfer occurs at the interface between
molecules in solid form, this means that there is no need for a liquid electrolyte.
81
It has been found, however, that interfacial properties are very important for
efficient solar cell devices using this architecture. Cells work best when the donor
and acceptor form separate domains that are roughly the same size as the exciton
diffusion length. There exist multiple ways to control these domains, from
choosing the appropriate spin-cast solvent, to annealing the device.68 The simple
methods of fabrication, ability to perform charge transfer without electrolytes and
efficiencies that are much higher than cells made out of single molecular
absorbers have spurred much interest in this technique.
4.4.2. Utilization of inorganic nanostructures
With ever-increasing knowledge of the construction of inorganic
nanostructures, some groups have begun to utilize them in order to increase the
efficiencies of organic photovoltaics. Many inorganic materials are excellent
electron acceptors which, when paired with complimentary organics, can increase
the rate at which excitons diffuse to electrodes. One way to incorporate inorganic
materials is to blend inorganic nanostructures with organic light absorbers, much
the same way that fullerene is blended with light absorbing polymers in the bulk-
heterojunction solar cell. In some cases, even the inorganic nanostructure can be
utilized with the organic to absorb light and capture more of the solar spectra.
Huynth et al. utilized this method of blending inorganic nanostructures
with organic materials by synthesizing CdSe nanorods in solution and mixing
them with the polymer P3HT.69 CdSe is a semiconductor material and its
absorbance spectra can be tuned by altering the bandgap of the structure which, in
this case, is accomplished by changing the dimensions of the nanorods. The
energy levels of the system are such that excited state electrons from the P3HT
can be readily transferred into the CdSe, while holes from the CdSe can hop to the
P3HT. Recombination is, therefore, more limited in such a system. Efficiencies
as high as 1.7% under Air Mass (A.M.) 1.5 Global solar conditions have been
achieved utilizing this method, again, without the need for liquid electrolyte.
Several other systems have been researched blending inorganic nanostructures
with organic light absorbers with
some success.70,71
In addition to being good
electron acceptors, some
nanostructures are believed to
actually alter the absorbance of
organic molecules. Metal
nanoparticles, specifically, have a
unique property known as surface
plasmon resonance. In the
presence of an electromagnetic
wave, the electrons in a metal
nanoparticle will oscillate with the
changing electric field (Figure 4.6).
This oscillation of charge around the surface of the nanoparticle creates its own
Figure 4.6. Schematic of metal nanoparticle in an electric field. Charges will oscillate as the electric field oscillate, creating a surface plasmon.
82
electromagnetic field as it becomes an oscillating dipole. The wavelength at
which the nanoparticle will absorb light strongly is given by the classical
electrodynamics expression for the polarizability of the nanoparticle with volume
V and relative dielectric constant of εr
2130 +
−=
r
rVεε
εα
4.1
Clearly, this quantity gets very large as the relative dielectric constant approaches
-2. For most metals, this occurs in the visible wavelength range. For instance,
gold nanoparticles appear very red in color because their plasmon absorption is
around 520 nm.
This re-radiation of incident light is what makes metal nanoparticles
potentially useful components in organic solar cells. As discussed, most organic
light absorbers have very low absorption cross sections (on the order of Å2)
which, coupled with the low rate of electron transport, typically hampers the
efficiency of organic solar cells. If metal nanoparticles can increase the intensity
of the electromagnetic fields, it is possible that they could increase the absorption
of organic molecules in these higher intensity fields. There have been several
reports of metal nanoparticles enhancing the fluorescence or organic
molecules72,73 and even some reports that metal nanoparticles increase the
absorption and efficiencies of organic photovoltaic cells.74,75 The physics of
surface plasmons and their effects on organic fluorophores are still not completely
understood however, and more research into what actually happens when an
83
organic light absorber is placed in close proximity to metal nanoparticles will no
doubt help illuminate ways to utilize these particles most effectively.
4.5. Porphyrin based molecular dyads and triads
Another interesting potential way to
circumvent the problem of recombination in organic
films is to create covalently bonded molecular donor-
acceptor assemblies that increase the molecular
excited state lifetimes. In this direction, much
research has been conducted examining the
possibility of utilizing porphyrins connected to
different acceptors and donors as photovoltaic elements. Figure 4.7 shows the
chemical structure of porphine, the main component of all porphyrins. In a
porphyrin, four pyrrole rings (cyclical groups that contain nitrogen) are joined
together by single carbon atoms to form a larger heterocyclic ring. The
conjugated nature of the molecule allows it to absorb visible light quite strongly.
A spectra of a porphyrin molecule (tetratolyl-porphyrin) is shown in Figure 4.8.
The strongest absorption band (around 416 nm) is known as the Soret band, and
the other bands between 500 and 800 nm are referred to as Q-bands. Porphyrins
are naturally found in several different biological processes. The unbonded
electrons in the nitrogens can ligate a metal ion, and this feature is utilized in
hemoglobin, which is a molecule that binds to iron in the blood. More relevant to
N NH
NNH
Figure 4.7. Chemical structure of porphine
84
solar cells, however, is the fact that porphyrins are utilized within chlorophyll
molecules as light absorbers in photosynthesis.
Since porphyrins are utilized as light absorbers in photosynthesis, it is
natural to believe that they could be
used to absorb light in an organic
photovoltaic. However, porphyrin
based photovoltaics suffer the same
problem of rapid exciton
recombination that all organic
photovoltaics suffer. In order to
reduce the rate of recombination,
some groups have researched ways
to covalently attach electron and
hole acceptors to porphyrins in order
to increase the lifetime of the charge separated state, hoping to extract the charges
more efficiently. A common electron acceptor utilized in these molecules is
fullerene. When attached to a porphyrin (see Figure 4.9) the assembly can be
referred to as a porphyrin-fullerene dyad. Due to the energy levels of the
porphyrin and fullerene (the excited electron in porphyrin is at 1.90 eV compared
to the lower 1.74 eV excited singlet state of the fullerene)76 the porphyrin, upon
absorbing light, readily donates its excited electron to the fullerene. When the
porphyrin is able to transfer the electron to the fullerene, the excited state lifetime
Figure 4.8. Absorbance spectra of tetratolylporphyrin
85
of the free base porphyrin-fullerene dyad extends from 20 ps to 290 ps.76
Further, a carotenoid can be added which accepts the hole from the porphyrin,
leaving the porphyrin neutral after donating an electron into the fullerene. This
type of structure is referred to as a molecular triad and the charge separated state
of a molecule like this can exist for up to 1500 ns.77
N
N
N
H N
N HN
C O O H
H
Figure 4.9. Chemical structure of a porphyrin-fullerene dyad molecule.
This fast electron transfer from porphyrin to fullerene (the rate constant is
1.8 x 1011 s-1 for the free base porphyrin-fullerene dyad in benzonitrile)76 mimics
observations in photosynthesis. Soon after chlorophyll absorbs light, the excited
state electron undergoes electron transfer from one compound to another,
preventing it from recombining so that the energy can be utilized by the cell. In
this way, photosynthetic organisms are able to extract energy from the sun and
turn it into useful chemical energy which, in one way or another, sustains almost
86
87
all life on the planet. Nature has clearly found a way to utilize solar energy
efficiently and mimicking this pathway may lead to advances in solar energy that
allow humans to utilize the sun for some of their present energy needs. Research
in molecular structures such as porphyrin-fullerene dyads may lead to
breakthroughs in solar cell technology, and potentially give insight into the
chemistry and physics of these photosynthetic pathways.
88
5. ILLUMINATED STM FOR SINGLE MOLECULE CONDUCTANCE
As discussed in the previous chapter, organic photovoltaics severely suffer
from recombination losses after the organic layer absorbs incident light energy.
Porphryin-fullerene dyads (see Figure 4.9) are currently being examined as
potential absorbing materials, due to their relatively long electronic excited state
lifetimes, which may prove to be a beneficial feature in photovoltaic applications.
Although there has been much research conducted on the spectroscopic properties
of these molecules,76 information on how these molecules conduct electricity at
the single molecule level, both in the dark and under illumination, may help to
shed light on how these molecules would perform in a photovoltaic device.
The STM break junction method (see chapter 1) is an ideal way to
measure the electrical behavior of these dyads at the single molecule level.
Furthermore, by outfitting the instrumentation with various optical components,
the electrical behavior of the molecules can be recorded under both dark and
illuminating conditions, allowing for the examination of the molecular
conductivity. To better understand these molecules, a laser-illuminated STM
system was constructed and the conductance of a porphyrin-fullerene dyad was
measured both in the dark and under monochromatic illumination within an
absorption band of the molecule. The following chapter outlines the
instrumentation for this system and concludes the results of measuring the
conductivity of a porphyrin-fullerene dyad.
5.1. Illuminated STM instrumentation
Figure 5.1 shows both a picture and a schematic of the illuminated STM
apparatus used to measure the conductivity of the porphyrin-fullerene dyads.
Several changes needed to be made to an existing scanning probe microscope to
incorporate an illuminating element in order to measure molecular conductivity
under laser illumination. In this section, the scanning tunneling microscope will
be described, along with the optical components needed to incorporate the
illumination and the environmental controls that needed to be modified in order to
keep the system in an oxygen-free environment.
Figure 5.1. Schematic and picture of illuminated STM system. On the left is a schematic representation and on the right is a photograph of the actual instrument.
5.1.1. STM
The STM system utilized in these experiments was a Agilent PicoSPM.
This scanning probe instrument has both AFM and STM capabilities.
89
90
Measurements involving illuminated CAFM (utilizing the same illumination
instrumentation) will be described in the next chapter. An STM scanner with a 1
nA/V current preamplifier was used for all dyad break junction measurements.
Initial scanning of surfaces was performed using Agilent PicoScan software. The
head electronics of the microscope were connected to a signal breakout box
(Agilent) for data recording and signal manipulation purposes.
5.1.2. Laser beam delivery
The samples (porphyrin-fullerene dyad monolayers on ITO) were
illuminated with an Ar/Kr laser (CVI Melles-Griot, Albuquerque, NM) which
could operate in the following wavelengths: 476 nm, 483 nm, 488 nm, 496 nm,
514 nm, 520 nm, 568 nm, 647 nm, 676 nm. The diameter of the beam from the
laser head was 0.7 mm, and the maximum power available at different
wavelengths was tens of milliwatts. The beam was launched into a single mode
optical fiber using a laser-to-fiber coupler with an achromatic lens for focusing
onto the 3.5 um fiber core (CVI Melles-Griot, Albuquerque, NM). The coupler
was adjusted to maximize power output from the other end of the single mode
fiber. Single mode fibers were utilized due to their small core diameters which
allows for smaller beam diameters after collimation from the other end of the
fiber. The use of single mode fibers leads to losses in power, but this does not
negatively impact experiments, as will be discussed in later sections.
The fiber was 2 meters long and was terminated in an adjustable fiber
collimator (Thor Labs, Newton, NJ). The fiber collimator contained an aspheric
91
lens with an estimated focal length of 4.6 mm which could adjust along the
optical axis. An adjustable collimator was chosen to correct for any changes in
the wavelength of the laser. The lens was positioned to collimate the light exiting
the fiber with the smallest amount of divergence in order to minimize error in
beam diameter and thus illumination intensity. At the sample the beam diameter
was 0.48 mm.
The collimator was screwed into a multi-axis lens positioner (Newport
Corporation, Irvine, CA) which was bolted down onto a small aluminum
breadboard (Thor Labs). The multi-axis lens positioner had adjustment in the x,
y and z direction. The x and y directions were used to locate the laser beam under
the STM tip. The z adjustment can be utilized in future experiments for focusing
the beam to tighter spots on the sample. A 0.3 optical density neutral filter was
placed in front of the multi-axis lens positioner for these measurements in order to
get a very low illumination intensity. A controllable optical shutter (Thor Labs)
was placed in front of the filter and could block the beam or allow the light to
pass through depending upon the applied voltage on the shutter (determined by a
separate controller).
Also attached to the aluminum breadboard was an environmental chamber
(explained in next subsection) into which the STM could be sealed. The laser
light was directed through the environmental chamber to a silver 45º mirror (Thor
Labs) which reflected the light to the sample. The sample holder was a standard
92
Agilent Pico STM sample holder, with a hole underneath the sample to allow the
reflected light from the mirror to reach the sample.
5.1.3. Environmental control
Porphyrin and other light absorbing organics can be very sensitive to the
presence of oxygen and can undergo irreversible photo damage. For this reason,
it was critical for experiments to be conducted in as oxygen-free an atmosphere as
possible. The experiments explained in this chapter were all performed in a
sealed environmental chamber, which had been purged with Argon gas prior to
recording data. To accomplish this, the glass from a cylindrical Agilent PicoSPM
environmental chamber was removed from its stainless steel base, and replaced
with a custom glass piece with a puckered hole. The hole (about 1″ in diameter)
was covered with an optical window (Thor Labs) where the laser light could
transmit through into the chamber with minimal scattering losses. A hole was
screwed through the stainless steel base in order to bolt the entire environmental
chamber onto the aluminum breadboard. A stainless steel ring was attached to the
top of the glass, and the STM could seal onto this ring with the sample and tip
securely sealed in the chamber. The base of the chamber contained gas inlet and
outlet valves that could be sealed once purging was complete. The entire
breadboard could be lifted and set onto a brick hanging from bungee cords in an
electrically shielded chamber to prevent both mechanical and electrical noise.
To further reduce the effects of oxygen in the tip-sample junction,
measurements were performed in liquid environment. Mesitylene, a
93
nonconductive solvent, was utilized for these measurements. Nonconductive
liquids are ideal for STM work since the metal tip (wire) is immersed in the liquid
during measurement. In order to utilize a conductive solvent, all but the very end
of the tip must be insulated with wax or polyethylene. The mesitylene was
bubbled with nitrogen or argon prior to use to remove any oxygen dissolved in the
solvent. The solvent was contained in a Teflon liquid cell held above the
substrate with spring clamps.
5.2. Illuminated single molecule experiment design
5.2.1. STM breakjunction experiment
The dyad samples were made prior to the experiments by immersing ITO
that had been cleaned with 5 minutes exposure to oxygen plasma in a solution of
5 uM porphyrin-fullerene dyad in mesitylene overnight and lastly rinsing with
mesitylene and drying in nitrogen. The porphyrin-fullerene dyads in this
experiment are shown in Figure 4.9. They contain a carboxylic acid on the
fullerene which can attach to the ITO, and a pyridine group on the porphyrin
which can attach to a gold STM tip. These molecules were prepared by the Gust-
Moore group at ASU.
The dyad STM breakjunction experiments were controlled by a labview
program and a National Instruments data acquisition device (NI USB-6221).
Measurements were taken at -0.4 V sample bias. The negative sample bias is
used in order to facilitate bonding of the pyridine to the gold STM tip (positive
relative to the sample). Picoscan software was first utilized to approach the tip to
94
the surface and an image of the surface was taken in order to determine the
quality of the tip and the surface. The Picoscan servo was turned off and the
labview program took control of the tip through an external z-position input on
the STM controller. The tip was held at a user-determined setpoint value above
the sample for a short period of time using a software PID controller which read
the current through the data acquisition device. The PID control was broken after
a set time and the tip was retracted and then re-approached to the surface where it
was held under PID control again. This process would be repeated multiple times
at several different places on the surface.
Data was recorded on a Tektronics TDS 2004B oscilloscope which
recorded current, piezo voltage, bias and optical shutter voltage. Once a set of
data for one withdraw was recorded on the oscilloscope it was immediately
transferred via USB cable to a computer where it was added to an open file
through another labview program. The entire process to withdraw the tip and
record the data took approximately 10 seconds, so a typical experiment of about 6
hours could collect as many as 2000 curves.
5.2.2. Illumination
Since the molecules are thought to be prone to irreversible photo damage,
it is important to understand whether any changes in conductance of the single
molecules are due to illumination or photo damage. A single sample, therefore,
needed to be measured in both the light and the dark during the same experiment.
To accomplish this, the optical shutter controller was set to automatically open
95
and close every ten minutes. To track which data were taken with the shutter
open and which data were taken with the shutter closed, a signal line from the
voltage output on the shutter controller was connected to the oscilloscope. Every
recorded curve also recorded the voltage on the shutter, which could be used in
data analysis in order to determine whether the sample was illuminated or not
during a given tip retraction. In this way, the same part of the sample can be
probed under illumination and dark conditions automatically, and the reversibility
of any changes in the molecules examined.
5.2.3. Data analysis
A labview program was created in order to analyze the break junction
data. Chapter 1 explains that when a molecule is trapped in the tip-surface gap
during withdraw, a step/plateau is observed in the current versus distance
retraction curve. When the current curves that have steps in them are compiled
into a conductance histogram, peaks appear that represent the molecular
conductivity of the molecules being studied. This works well with very simple
molecules, like alkanethiols, on a gold surface with a gold tip. Conductance
peaks in such a system are very clear and easy to characterize as can be seen in
Figure 1.12.
The gold tip-dyad-Indium Tin Oxide system, however, is a much more
complicated experiment. Unlike the gold-thiol bonds, carboxylic acid-ITO bonds
are not as well defined. Changes in bond formation can lead to heterogeneity in
the conductance of the molecules. This can lead to wide conductance peaks that
96
are difficult to interpret if the analysis is completed by simply compiling all
curves that have plateaus. For this reason a different data analysis program was
utilized in order to better interpret the data.
One way to differentiate peaks from noise is to remove the sloping
background from conductance histograms. The curves are essentially exponential
decays, so a current versus time curve has more data points at low current. When
these curves are compiled into a single histogram, it leads to significant sloping
effect, where very low conductance values have very high counts in the
histogram. This can make it difficult to read peaks that may be lower in
conductance. In order to remove the background of the histograms, it is
beneficial to only compile the points within the step/plateau, and not the rest of
the curve.
This step analysis was completed with a labview program that created a
“window” of a user-specified number of data points that was automatically “slid”
over the entire decay curve, point by point. In order to determine if a step
occurred within the window of data, a histogram of the data in the window was
produced. Since this histogram is also a histogram of a decay curve, it decays in a
similar way as the entire decay curve. The histogram is then automatically
analyzed by the software. The maximum and minimum count values in the
window are determined. If the difference between the maximum value in the
window and the minimum value in the window is determined to be above a user-
selected threshold, and the maximum occurs in the middle of the window, then
97
the position of the maximum value is assigned a single count. This procedure
eliminates the exponential decay background that exists when creating a
histogram from complete curves, since the count will only be assigned if the
maximum occurs in the middle of the window. This produces histograms that
have zero counts at zero conductance, and peaks in the conductance histograms
are much more readily analyzed. This procedure has been previously reported in
determining the conductance of oligoalanine.78
5.3. Illuminated single molecule conductance results
5.3.1. Conductance of illuminated dyads
Figure 5.2. Conductance histograms of porphyrin fullerene dyads on ITO. The red histogram is with the light off, and the blue histogram is with the light on. The insets in both pictures are light on and off histograms of experiments conducted on blank ITO samples.
The conductance of the porphyrin-fullerene dyad, both in the dark and
under laser illumination, was determined using the instrumentation described in
the previous section. Since the dyad molecules contain a porphyrin, it is believed
that the conductance of the molecule may be affected by photon absorption of that
porphyrin. The molecules were exposed to 520 nm laser illumination for 10
minutes at a time, followed by 10 minutes of darkness as controlled by the
98
99
programmable optical shutter. This allowed for the reversibility of any changes in
conductance to be determined. Figure 5.2 shows the results of the experiment
performed at a laser intensity of 183 ± 13 mW/cm2. The conductance histogram
in red shows the conductance values of the molecule during the dark periods,
while the blue histogram shows the conductance values of the molecule during the
illuminated periods. The peak in both histograms between 0-1 nS is an artifact of
the measurement and/or the data analysis and occurs in control experiments with
clean samples (Figure 5.2 insets).
In examining the dark conductance histogram, the molecule exhibits a
singular conductance peak, which is determined to be the single molecule
conductance of the dyad in the dark. This peak is centered at 2.45 nS. The
illuminated histogram, however, looks remarkably different. The peak around
2.45 nS still appears, but is much broader, and another peak around 4.3 nS is also
apparent. This second peak goes away quickly in the absence of light, since the
non-illuminated histogram does not exhibit this conductance feature. This is clear
evidence that the conductance of a portion of the molecules changes in a
reversible manner upon illuminating at 520 nm.
This illuminated experiment was completed at different laser intensities to
determine if any relation existed between the flux of photons on the surface of the
ITO and the monolayer, and the strength of the second peak seen in Figure 5.2.
The two peaks shown in the illuminated histogram were non-linearly fit to two
Gaussian peaks, and the positions, amplitudes and widths of those peaks was
determined. The first peak is labeled P1 and the second peak is labeled P2. The
two peaks were compared with the following formula
1
2
)*(
)*(
PFWHMAmplitudeGuassian
PFWHMAmplitudeGuassianRatio =5.1
This ratio was plotted against the laser power density in mW/cm2 and is
shown in Figure 5.3. Clearly, as the power density increases, this ratio increases
nearly linearly until a peak power density around 270 mW/cm2 is reached, at
which point the ratio levels off. So, as the power density increases, the area under
the Gaussian curve of the second peak compared to the first peak also increases,
signifying that more of the molecules are exhibiting higher levels of conductance.
0 1 0 0 2 0 0 3 0 0
0 .0
0 .4
0 .8
1 .2
Rat
io o
f P2:
P1
L a s e r P o w e r D e n s ity a t 5 2 0 n m ( in m W a tt /c m 2 )
Figure 5.3. Graph of peak ration versus laser intensity. The higher conductance peak becomes stronger as the laser intensity increases.
100
Since the ratio levels off around 270 mW/cm2, higher laser intensities were
unnecessary in the apparatus. It is possible that higher intensities were
accelerating molecular photodamage.
5.3.3. Free-base porphyrin
The conductance of a free base porphyrin with both carboxylic acid and
pyridine functional groups was also measured under illumination. This molecule
was chosen due to its lack of fullerene attachment in order to determine whether
or not the fullerene plays a role in the change in conductance upon illumination.
The conductance histograms of this molecule (not shown) do not indicate a higher
conductance peak upon illumination, despite repeated experiments at numerous
laser illumination powers.
5.3.4. Control experiment
3-(pyridine-4-yl)-
propionic acid was utilized as a
control molecule. This molecule
is a short alkane chain that has
the same functional groups
(carboxylic acid and pyridine) as
the dyad molecule used in this
experiment. However, this
control molecule is not optically
active. Figure 5.4 shows the
0 10 20 30 400.0
0.4
0.8
0 10 20 30 400.0
0.4
0.8
Light Off
Light ON
Cou
nts
(Arb
Uni
ts)
Conductance (nS)
Figure 5.4. Light on and off histograms of 3-(pyridine-4-yl)-propionic acid
101
102
results of this experiment in the dark (red) and in the light (blue). The molecule
shows little to no difference in conductance upon illumination, which is the
expected result due to the lack of optical absorption at 520 nm.
5.3.5. Conclusions
The new conductance peak observed after illuminating the molecules with
520 nm is a remarkable feature. The first thought would be that the feature
represents increase in conductance due to photogenerated charges. However, the
power intensity utilized in this experiment is not sufficient enough to actually
cause a noticeable shift in molecular conductance due to photogenerated charge.
Even if every single photon that hit the molecule being probed was absorbed and
the excited state charge was transferred into the electrodes generating a
photoinduced current, that photocurrent would be much less than a picoampere,
which is less than the detection limit of the experiment. Taking that into account,
the substantial increase in conductance must be due to a factor other than
photoinduced carrier generation.
The increase in conductance seen in the illuminated histograms has been
ascribed to changes in the molecule during excitation that change how the
molecule conducts electricity. The natural change to explore in this system is the
electron transfer from porphyrin to fullerene. As explained, in solution, once the
porphyrin absorbs light the excited state electron immediately transfers into the
fullerene electron acceptor, extending the excited state lifetime of the molecular
unit. This charge separated state has a different electronic configuration than the
ground state of the molecule, and may very well be the source of the change in
molecular conductance. Evidence for this separated-state theory can be seen in
the illuminated conductance of the free-base porphyrin without the fullerene.
This molecule does not have a higher illuminated conductance, which aligns with
the idea that the fullerene is, therefore, a necessary component in the molecular
dyad conductance change.
This theory was tested with optical spectroscopy. Transient absorption
was performed on ITO samples with dyad layers using a 520 nm laser line to
pump the samples. The absorption kinetics between 550-880 nm (where
absorbance can be attributed to
porphyrin radical cations) show
a very long lived state that went
out to 14 us. Further, a transient
absorption spectra showed the
appearance of an absorption
state around 640 nm. Figure 5.5
shows the transient absorption
spectra of the dyads on ITO with
the state at 640 nm appearing clearly after 3 ps (red squares), decaying after 10 ps
(green circles) and almost, but not completely, gone after 2 ns (blue triangles). A
similar spectra of free-base porphyrin on ITO shows no appearance of this peak.
Figure 5.5. Transient absorption spectra of porphyrin-fullerene dyads on ITO. The red squares are after 3 ps, the green circles are after 10 ps, and the blue triangles are after 2 ns.
103
104
This peak is attributed to the formation of the porphyrin radical cation in the
porphyrin-fullerene dyad on ITO. The long lived state (up to 14 us) shows that
this cation is sometimes very long lived within the dyad film. Further, the fact
that this state goes away very quickly after the light is turned off (almost
completely gone after 2 ns) explains why the conductance goes back to the
completely dark conductance after the light is turned off in the STM apparatus.
This might also explain why the ratios of the two histogram peaks (Figure 5.3)
changes with laser power intensity. There are probably more molecules in the
cationic state when the laser intensity increases, increasing the number of high-
conductance species.
The measurements on dyad molecules show how the scanning tunneling
microscope may be utilized to understand how molecular conductance of novel
molecules can be optically controlled. The ability to study this molecular
arrangement at the single molecule level was necessary in order to observe this
effect, due to fact that there are two very clear populations of molecules that have
different conductances. A bulk measurement, by contrast, may have yielded an
average measurement that does not fully explain the phenomenon. The STM was
the ideal measurement technique for these molecules, and the instrumentation
changes were necessary and vital in understanding illuminated porphyrin-
fullerene molecular conductance. This measurement, in particular, shows the
versatility of scanning probe instrumentation.
105
6. PORPHYRIN-FULLERENE DYAD THIN FILM GROWTH AND
ELECTRICAL CHARACTERIZATION
It was recently found that the molecule in Figure 4.9 could organize to
form multiple layers on an Indium Tin Oxide surface after the porphyrin ring had
been protonated. The first evidence of this phenomenon came after sonicating the
molecule in 1,2,4-Trichlorobenzene. The sonication of the solvent releases
chlorine atoms, forming hydrochloric acid in the solvent. This acid, although
very low in concentration, is enough to protonate the, typically low, concentration
of the dyad in solution. A film of the protonated dyad molecules would then grow
on an ITO substrate from this sonicated solution, thicker than a self assembled
monolayer. It was determined that protonation of the dyad leads to these layers
forming, as unprotonated dyads do not form such films. As a test of the
illuminated scanning probe system, these protonated films were measured under
both illuminated CAFM and illuminated mercury drop electrode apparatus for
comparison purposes.
6.1. Film growth and characterization
As will be mentioned in the section describing CAFM measurements on
these films, reliable data of photocurrent and photovoltage was only actually
obtained for mercury drop experiments. For this reason, the materials and
methods utilized in the mercury drop experiment are much more uniform since
the best procedures for making films could only be determined after repeated
photovoltaic results. In the materials and methods section, the set protocol for the
106
mercury drop experiments are noted, as are the different general methods for
producing samples for the CAFM experiments.
6.1.1. Materials and methods
Sonicating solutions of dyad in 1,2,4-Trichlorobenzene is a somewhat
haphazard way of protonating the molecules. For this reason, once it was
determined that protonation caused dyad film growth, the organic acid
Trifluoroacetic acid (TFA) was utilized as the protonating agent, in lieu of the
sonication to produce HCl in solution. For a solution of 5 uM, a single drop of
TFA is enough to protonate the entire solution, however multiple drops were
typically utilized to ensure that the solution remained protonated throughout
sample deposition, as TFA is more voltatile that 1,2,4-Trichlorobenzene and can
evaporate out of solution easily.
In order to make sample solutions, 1,2,4-Trichlorobenzene was first
stirred with sodium bicarbonate in order to remove residual HCl that might exist
in the solvent as bottled. This allows for TFA to be the only protonating agent in
the sample solutions. For CAFM experiments, the 1,2,4-Trichlorobenzene was
distilled once under reduced pressure prior to use. For mercury drop experiments,
it was found that distilling twice gave better photovoltaic results, probably due to
the removal of residual sodium bicarbonate left in the solvent after the first
distillation. The distillation apparatus was cleaned with distilled water and
ethanol between distillations for the mercury drop experiments. 1,2,4-
107
Trichlorobenzene was stored after distillation in the dark to prevent any HCl
forming in the solution and was purged with Nitrogen prior to use.
For all experiments, a stock solution of the dyad molecule was prepared
by stirring several milligrams of the molecule into about 15 mL of 1,2,4-
Trichlorobenzene until dissolved. For mercury drop experiments, a couple drops
of TFA was also added to the solution prior to stirring, and the solution was
heated. The TFA was found to help the dyad dissolve into the solvent better. For
mercury drop experiments, the solution was allowed to cool before further testing.
The concentrations of stock solutions were checked by absorbance in a
UV-Vis spectrophotometer (Varian, Palo Alto, CA). Spectra were taken from
200-800 nm. For the mercury drop experiments, stock solution concentration was
determined in protonated form, whereas this varied in CAFM experiments. In the
mercury drop experiments, the spectra was also utilized to determine if the
solution was fully protonated by the TFA. Once concentration of the stock
solution was determined, it was diluted down to the desired sample solution
concentration, which was usually 5-10 uM for CAFM experiments and was
always 10 uM for mercury drop experiments. A couple more drops of TFA was
typically added during the dilution process to maintain proper dyad protonation
throughout sample deposition.
Background UV-Vis spectra of the ITO slides on glass (Delta
Technologies) was typically taken in the UV-Vis spectrophotometer prior to
sample cleaning. Samples were cleaned by sonicating in acetone for 10 minutes
108
to remove large residual particles, and then cleaned for 5 minutes in oxygen
plasma to remove any leftover organics from the surface of the ITO. ITO slides
were then immediately removed from the oxygen plasma and submerged in the
sample solutions. The solutions were then purged by bubbling with nitrogen
through a cleaned needle, and sealed under nitrogen with parafilm. Samples were
left in the solution in the dark for a determined length of time. For CAFM
experiments, samples were left anywhere from 1 day to 2 weeks, while, for
mercury drop experiments, samples were left to soak in solution for 6 days.
ITO slides removed from sample solutions were typically green
immediately after removal After removal from solution, CAFM samples were
rinsed with both 1,2,4-Trichlorobenzene and toluene and dried in nitrogen. TFA
was sometimes added to these solvents to maintain the protonated state of the
films. Mercury drop samples were dried in a vacuum oven for at least an hour at
50 C. Both rinsing with solvent (without TFA added) and drying in the vacuum
oven caused the color of the films to change from bright green to yellow,
indicating a deprotonation of the porphyrin following removal from the sample
solution, which was probably a result of some TFA evaporation.
6.1.2. Spectral properties of dyad thin films
All spectral measurements were taken with a UV-VIS spectrophotometer
(Varian, Palo Alto, CA). Background absorbance of molecular solutions were
taken with clean solvent in a 1 cm pathlength quartz cuvette, and background
absorbances of ITO were taken by placing clean ITO slides in the cuvette holder
Figure 6.1. Absorbance spectra of 5 uM free-base and protonated porphyrin-fullerene dyad in 1,2,4-Trichlorobenzene.
in the instrument. Figure 6.1 shows the absorbance spectra of the dyad in
solution. The black circles represent the spectra of a solution of 5uM free-base
(not-protonated) dyad in 1,2,4-trichlorobenzene. The blue squares represent the
spectra of the same solution after protonation with TFA. After protonation in
solution, the soret peak shifts from 415 nm to 445 nm. The full width at half
maximum of the Soret also changes from about 15 nm to 30 nm, indicating
absorption broadening of the Soret after protonation. The absorbance at 650 nm
(Q-band) also increases after protonation. The color of the solution visibly turns
from yellow to green after protonation with TFA.
Figure 6.2 shows the absorbance spectra of a monolayer of dyad on ITO
created from a 5 uM solution of the dyad in TCB, and a spectra of a dyad film
grown from a 5uM solution of protonated dyad in TCB. The deposition time for 109
110
both samples was 5 days. The Soret peak for the sample made with unprotonated
dyad was at 430 nm, while the peak position for the sample made from protonated
dyads was at 435 nm. The Soret absorbance for the sample made from un-
protonated dyads was 0.062, while the soret-absorbance of the sample made with
protonated dyads was 0.32 and the ratio of the FWHM of the two soret peaks was
roughly 2:3 (actual values were 28 nm for the sample made from unprotonated
dyad and 41 nm for the sample made from protonated dyad), indicating that the
sample made from the protonated dyad also showed peak broadening. The red-
shifted peak of the sample made from protonated dyads, along with the peak
broadening and the fact that the peak at 650 nm is also relatively larger than the
other Q-band peaks indicates that, after removal from the solution, not all of the
film de-protonated back to the free-base form of the porphyrin. However, since
the overall shape of the spectra of the sample made with protonated dyad has the
features that are associated with the free base sample, it is believed that the
majority of the porphyrin in the film reverted to free-base form.
Figure 6.2. Absorbance spectra of porphyrin-fullerene dyad on ITO
6.2. Attempts at CAFM photocurrent measurements
6.2.1. Materials and methods
111
Numerous attempts were made to study photocurrent properties of dyad
thin films on ITO with illuminated CAFM. The following outlines a typically
CAFM experiment on dyad thin films. The same illuminated scanning probe
instrumentation that was outlined in chapter 5 was employed to study these films
with the AFM instead of the STM. The only major differencse with the AFM
system was the use an AFM scanner in the microscope and the use of an AFM
photodetector with an optical bandpass filter positioned over the photodiode. This
bandpass filter allowed for the AFM laser light to reach the photodiode, but
blocked out the green illuminating light from the Ar/Kr laser. Samples were
112
prepared as outlined in the previous section and were tested in multiple solvents,
including 1,2,4-Trichlorobenzene, toluene and toluene with TFA added. AFM
tips were coated with chromium and gold in consecutive layers. This procedure
was outlined in chapter 3, however, for many of the CAFM experiments more
layers were utilized. Conductive diamond tips (Mikromash) were also used in
some CAFM experiments. Illumination was conducted at 520 nm at different
powers.
The current, shutter voltage and deflection were directly wired through
BNC cables to the National Instruments data acquisition system (USB-6221),
while the piezo voltage was routed through a homemade voltage divider and
follower before connection to the data acquisition system. These four values were
monitored and recorded with custom-written labview software. The AFM tip was
approached to the surface through the PicoScan control software to a user-
specified tip deflection (typically 1 V on the AFM photodiode). Once the tip had
settled on the surface, the labview program was started and the shutter was
programmed to open and close at set times. The labview program recorded the
photocurrent as the shutter opened and closed, along with the deflection (system
was usually kept under deflection control), shutter voltage and piezo voltage.
6.2.2. CAFM photocurrent results and challenges
Figure 6.3. CAFM photocurrent measurements. The top graph shows the current as the light is switched on around 70 seconds and off at 130 seconds. The current increases to about 6 nA when the light is turned on. The lower graph shows the voltage applied to the piezo (divided by 10 through a voltage divider). Clearly, when the light is turned on, this voltage changes, reflecting a disturbance in the AFM tip. The voltage bias was 0.005 V.
Figure 6.3 shows a typical result from a CAFM photocurrent experiment.
Although, it appears as though the current is increasing rapidly when the sample
is illuminated (70 seconds), changes in other signals indicate that the increase in
current may be related to more than light absorption by the dyad film. The
voltage on the piezo (also shown in Figure 6.3) also changes when the sample is
illuminated. Since the servo is on in this experiments, the piezo voltage changes
with changes in the deflection of the AFM tip. Typically, when an AFM tip is on
the surface, these changes are quite small and the piezo voltage goes up and down
113
to keep the cantilever at a stable deflection. However, as can be seen in Figure
6.3, the change in piezo voltage is not random and occurs precisely as the shutter
opens and closes. This seems to indicate that there is a change in cantilever
deflection upon opening and closing the shutter. If this is true, then the light may
be inducing a change in the force between tip and surface, leading to changes in
current which may be unrelated to photon absorption by the film.
This was
explored further by
examining the effect
that laser illumination
has upon tip deflection
by removing the tip
from sample surface.
The deflection of the
cantilever was
monitored away from a
surface both before and
after illuminating it
with the Ar/Kr laser. Figure 6.4 shows the results of the change in tip deflection
versus cantilever spring constant upon laser illumination. Clearly, as the spring
constant decreases, the amount that the AFM laser deflects gets larger. This
suggests that the laser can actually induce a bending on the cantilever, the
0.05
0.1
0.15
0.2
0.25
0.3
0 1 2 3 4 5 6 7
Cantilever Bending Upon Ilumination
Volta
ge C
hang
e R
ead
on P
hoto
diod
e (b
endi
ng)
Cantilever Spring Constant (N/m)
Figure 6.4. Effect of laser illumination on cantilever bending. As the value of the cantilever spring constant gets larger, the voltage change read on the detector gets smaller, indicating less bending of the cantilever upon illumination.
114
displacement of which is inversely proportional to the spring constant. This, of
course, suggests flaws in the experimental design if the probe is bending due to
interactions with the illuminating laser. Changes in contact forces can, by
themselves, lead to changes in the current recorded by the CAFM.
Other methods
were employed to reduce
the effect of this
interaction between the
probe and the light. In
one method, the tip was
first approached to the
surface either in the light
or in the dark. IV curves
were then measured in the
light and in the dark, and
the system was
maintained in a static illuminating condition throughout the measurement.
Influences of the light on the cantilever were then reduced. Initial measurements
on porphyrin thin films (made in a similar method to the dyad thin films) utilizing
a conducting diamond tip were promising, as the IV curves (Figure 6.5) showed
significant open-circuit voltage as well as fill factor, defined as the maximum
power output divided by the theoretical maximum determined by the short circuit
Figure 6.5. IV curves of porphyrin layer made with conductive diamond AFM tips. The blue curve, while not showing as much overall current, shows significant photovoltage as evidenced by where the curve crosses the 0 current axis.
115
116
current and the open-circuit voltage. The conductive diamond was thought to
enhance the photovoltaic properties of the system due to having a favorable work
function that created a larger energy gap than the typical gold tips, However,
when these measurements were repeated on a blank ITO slide, similar
photovoltaic results were obtained. It is thought that the boron doping in the
conductive diamond tip makes for a p-type semiconductor. When in contact with
ITO, then, the tip can form a P-N junction with the n-type ITO, and work as a
photovoltaic device without the need for another absorbing material. IV curves
on dyad layers with a gold tip showed no measurable photocurrent or
photovoltage.
6.2.3. Calculations of expected photocurrents
Since the AFM seemed inadequate in determining photocurrent from dyad
films, it was necessary to calculate what the theoretical maximum photocurrent
would be given the dimensions of the tip and the absorptive properties of the film.
This helped determine whether the AFM system was not working properly, or
whether the system was simply incapable of observing even the maximum
possible generated photocurrent given the AFM specifications. This was
accomplished by first examining the absorption spectra of a typical dyad film on
ITO (Figure 6.2).
The extinction coefficient of the porphyrin-fullerene dyads in solution at
the Soret peak is about 380,000 M-1cm-1. The width of the peak on the sample,
however, was approximately 1.75 times the width of the peak in solution. So,
using a guassian fit for the Soret peak, the extinction coefficient of the molecule
on the ITO sample can be estimated at approximately 220,000 M-1cm-1. The peak
absorption of the film in the absorption spectra (Figure 6.2) is approximately
0.375. The concentration of the molecules on the sample can then be given by
Beard’s law as
clAbsorption ε= 6.1
Where ε is the extinction coefficient, c is the concentration and l is the pathlength.
If, instead of being on a slide, the molecules were in a 1 cm pathlength cuvette,
the concentration would be approximately 1.7 uM or 1.02 x 1015 molecules/mL in
the cuvette. If, now, those molecules were condensed onto two 1 cm x 1 cm
surfaces (the front and back of the ITO slide), that would mean that each side of
the ITO contained 5.12 x 1014 molecules. The molecular area had been
previously determined by molecular modeling of tetraphenyl-porphyrin as 213 Å2.
Dividing the area of the slide probed by the spectrophotometer by this area gives
about 5 x 1013 molecules per layer on the surface. Dividing the number of
molecules by this gives approximately 10 layers of molecules on the surface of
the slide.
A power intensity that has been shown not to damage the molecules in the
breakjunction experiment is approximately 0.183 W/cm2. Converting this to
photon flux per Angstrom squared (the absorption cross sectional area of the
molecules) gives
117
s
PhotonscmsmW
photonsxcm
mW
*55
)10(
)100(**
103*1832
210
215
2 °°
Α=
Α
6.2
The area of the AFM tip contact is calculated by Hertz theory
3269
993 102
10*10)10*1(*)10200(* mx
PaNmx
KFRa −
−−
===6.3
Where a is the radius of the circular contact area, R is the radius of the probe, F is
the force of contact and K is the Young’s modulus of the material (estimated here
as 10 GPa).79 Thus, the area of the contact is approximately 23 nm2. The area of
the molecule, as mentioned above, is approximately 2.13 nm2, and so the number
of molecules being probed by the AFM tip is then
108layers 10*13.2
232
2
≈=nm
nmN 6.4
Since every molecule has an absorption cross sectional area of approximately 1
Å2, then the photon flux per Å2 can be used to determine the number of electrons
generated by the film in contact with the AFM probe, per second.
pAxs
electrons
s
42 1052.95940)*
Photons 55(*molecules) (108 −
°==
Α
6.5
This calculated photocurrent is well below the detection limit of the CAFM
preamplifier. Even significant changes to the quantities in this calculation (lower
elastic modulus or larger tip radius) do not change the final result enough for the
sensitivity of the current instrumentation. Therefore, it was decided that the
CAFM is an ineffective instrument for studying photocurrent behavior in these
samples. This is, of course, compounded by the various artifacts found when
118
119
doing the actual measurements, such as laser interference and tip doping
complications.
6.3. Mercury drop photocurrent measurements
Clearly, based upon the calculations, the AFM tip is simply too small to
measure any photovoltaic effects with the current instrumentation. In order to test
this assumption, and to see if these dyad films truly had any photovoltaic
properties, a mercury electrode was also utilized to study these films under
illumination. The size of the mercury drop is much larger than the size of the
AFM tip, and can be controlled to a greater degree by using different drop
volumes and contact areas. The size of the electrode contact area can also be
easily determined by optical techniques, as opposed to the AFM tip which
requires electron microscopy in order to determine tip radius.
6.3.1. Materials and methods
Figure 6.6. Mercury drop electrode instrumentation. A) A photograph of the entire setup with CCD camera positioned over the beam splitter and monochromator to the left. B) A close up of the environmental chamber with beam splitter above.
Samples were prepared for mercury drop experiments as described in the
previous sections. The average peak absorbance for the samples measured was
0.34 ± 0.10. Figure 6.6 shows a picture of the mercury drop experimental
apparatus. All experiments were conducted in an aluminum environmental
chamber to prevent electrical noise in the measurements. The sample was
attached ITO-side down to the sample holder and made electrical contact with a
piece of brass. This piece of brass was then in electrical contact with a Kiethley
electrometer outside of the aluminum environmental chamber. Light from a
Xe/HgXe lamp ( Oriel) and monochromator (SPEX) was diverted to the sample
by reflection off of a 50/50 beam splitter situated on top of the environmental
120
121
chamber and through a window placed in the chamber lid. Part of the beam was
transmitted through the beam splitter to a laser power meter also situated on top of
the environmental chamber.
At the bottom of the environmental chamber, a stainless steel precision
adjustment screw was attached which translated up and down within the
environmental chamber. The drop of mercury could be situated on top of this
adjustment screw and then be twisted up to be in contact with the Indium Tin
Oxide. The precision screw was grounded, effectively grounding the mercury
drop. The sample holder within the environmental chamber was tension fit onto
the non-rotating spindle of a micrometer whose adjustment knob screwed into the
side of the chamber. This allowed for the sample to be moved in one dimension
so that multiple locations on the same sample could be probed by the mercury
drop.
Before testing, the intensity of the illumination was determined. A
calibrated photodiode was placed at approximately the same position as the
sample. The sample holder and the beam splitter were positioned until the desired
intensity was achieved. Once this had been completed, a spectra of the lamp was
recorded by moving the monochromator from 300 nm to 800 nm and observing
both light and dark currents at wavelengths in 5 nm increments. This allowed for
future action spectra to be calibrated in order to determine the incident photon-to-
electron conversion efficiency (IPCE).
122
A CCD camera with telescope zoom was situated above the beam splitter
looking down through the window into the environmental chamber. The camera
was able to focus on the contact area of the mercury drop on the ITO. For every
spot tested on the film, a picture of the drop on the ITO was recorded. The
contact area diameter was calculated by determining the diameter of the precision
screw which was also in the camera frame. By determining the diameter of this
screw with a precision micrometer, the diameter of the contact could be directly
read from the camera image.
The 20 uL mercury drop was placed in electrical contact with the precision
screw by dropping it on the top of the recessed tip of the screw. The drop could
then either be used for experiments as a clean electrode, or a monolayer of
thiolated organic molecules could be self assembled on the mercury drop.
Various experiments done previously by other groups have shown improved
photovoltaic performance with the addition of an interfacial layer between
electrodes and the photoactive material.80,81 One reason that an interfacial organic
layer may improve performance is the idea that an organic with an intrinsic dipole
moment may direct current in a diode like way, forcing electrons or holes to flow
in a singular direction and preventing recombination with the photoactive material
after carriers have percolated to the electrodes. This could be especially
important with the mercury apparatus, as the work function of mercury is very
similar to that of ITO (4.49 eV for mercury and 4.3-4.75 eV for ITO).82-84 This
means that the electrodes by themselves do not provide a strong current direction
123
preference. In these experiments, a hexanethiol monolayer was utilized in order
to determine if it could improve upon the photovoltaic properties of the dyad
films. Previous mercury drop experiments have shown that alkanethiols can
exhibit rectifying behavior when sandwiched between a semiconductor substrate
and a mercury electrode.85 It is then believed that the hexanethiol may prevent
electron-hole recombination in the dyad film.
Hexanethiol monolayers were assembled on the mercury drop by applying
a drop of neat hexanthiol solution directly to the mercury drop sitting on top of
the precision screw. Afterwards, the experimental chamber was sealed and
flushed with nitrogen for five minutes. A kim wipe was utilized to remove any
excess hexanethiol on the mercury drop, and the chamber was resealed under
nitrogen for another 5 minutes. Hexanethiol is very volatile so after 5 minutes in
a nitrogen atmosphere any unattached molecules evaporate off of the mercury
drop leaving the self assembled monolayer ready for testing.
Figure 6.7. CCD image of mercury in contact with dyad thin film sample. The inner circle is the mercury contact, while the outermost circle is the edge of the precision screw.
After contact with the ITO sample, a picture of the mercury drop was
taken with the CCD camera positioned over the optical beam splitter. A photo of
a typical mercury drop contact area is shown in Figure 6.7. The center circle is
the mercury in contact with the ITO, while the largest circle is the edge of the
precision screw that supported the mercury drop. This outer circumference was
utilized to calibrate the area of the mercury drop contact.
124
Electrical tests were performed to determine the photovoltaic properties of
the dyad thin film. The first test was a photocurrent experiment completed under
white light illumination. The current between the ITO and the mercury drop was
recorded on the Keithley electrometer at 0 bias. After a given time the sample
was illuminated from reflection off of the beam splitter for a time, before the light
125
was blocked from the sample again. Similarly, a photovoltage measurement was
recorded with the same technique, but with the electrometer recording the voltage
between the two electrodes instead of current. An IV curve was also recorded in
the dark and in white light illumination. Finally, an action spectrum of the
photocurrent was completed by ramping the monochromator from 300 nm to 700
nm, with 5 nm increments. For each wavelength, the dark current was recorded
along with the illuminated current. The subtraction of the two would then give
the photocurrent generated at each wavelength by the dyad sample.
6.3.2. Mercury drop electrical results
Initially, 6 samples were prepared as described above in the materials and
methods section for dyad thin film growth. Three samples were tested with a
mercury drop that had been treated with neat hexanethiol solution as described
above, while 3 samples were tested with a bare mercury drop. All experiments
were performed in a nitrogen atmosphere. 5 spots on each sample were tested,
and the positions of each spot were recorded on the micrometer. The same
horizontal positions were tested on all six samples.
Figure 6.8
shows a typical
photocurrent
measurement taken
with a mercury drop
with a hexanethiol
monolayer. The
current is negative
which indicates that
the electrons are
going into the
mercury drop. Almost all spots with the hexanethiol monolayer show some
transient photocurrent with white light illumination, whereas, with the bare
mercury drop, only 6/15 spots show any measurable photocurrent. This may be
due to the fact that the current in the dark at 0 V is often over 1 uA for the bare
mercury drop, making it difficult to detect small changes in current on the order
of nanoamperes. The lack of measurable photocurrent may also be due to the lack
of current direction preference explained earlier. The average photocurrent
density measured with the hexanethiol drops was -1.56 ± 1.42 uA/cm2, while the
average photocurrent density measured with the bare mercury drops was -1.32 ±
0.96 uA/cm2 (only included were drops that showed measurable photocurrent).
-6
-5
-4
-3
-2
-1
0
0 4 8 12 16 20 24 28
Dyad Thin Film Photocurrent Density
Cur
rent
Den
sity
(uA/
cm 2
)
Time (s)
Figure 6.8. Dyad thin film photocurrent density measurement.
126
Figure 6.9 Average photocurrent density versus drop position, measured from position 1. The mercury drop had a hexanethiol monolayer. Spot 1 only had measurable photocurrent on 1 sample, so there are no error bars a the zero position.
Figure 6.9 shows the average value of the photocurrent density at each of
the 5 spots with the hexanethiol drop. Values were only taken if the photocurrent
density was measureable and are noted as such. The samples were only 1″ long,
and it is noted that the values of the photocurrent density uniformly increase
towards the middle of the sample, and then decrease when approaching the other
127
end of the sample. This could be due to non-uniform film thicknesses depending
upon position, or non-uniform ITO electrical properties.
0
0 .5
1
1 .5
2
0 5 1 0 1 5 2 0 2 5
D y a d T h in F i lm P h o to v o lta g e
Volta
ge (m
V)
T im e (s )
Figure 6.10. Dyad thin film photovoltage measurement.
Figure 6.10 shows a typical photovoltage measurement taken with a
hexanthiol drop. The photovoltages measured for all drops were relatively low,
with the largest photovoltage response measuring about 1.75 mV. The average
measured photovoltage for all hexanethiol drops and positions was 0.20 ± 0.45
mV. The hexanthiol drops, again, showed larger values of photovoltage than the
bare mercury drops (average photovoltage for bare mercury drops was 0.12 ± 0.18
mV) which could be due to the reduction in charge recombination with the
hexanthiol layer. The large standard deviations seen in both the hexanethiol and
the bare mercury drops are due to variations in measured quantities based upon
128
129
position of the drop on the sample. This variation in position is similar to the
photocurrent density measurements.
Due to the presence of artifacts in the CAFM photocurrent data, it was
important to obtain evidence that no similar artifact occurred in the mercury drop
experiments which would lead to a false photocurrent observation. Specifically, it
was important to validate that the photocurrent was due to the presence of the
dyad film on the ITO and not some heating effect or effect between the mercury
drop and the ITO (as was the case with the conducting diamond AFM tip). If the
photocurrent occurred in wavelengths that the dyad absorbs strongly, then this
would be strong evidence that the photocurrent occurs due to the dyad film. As
such, the photocurrent action spectra gave reliable evidence that the photocurrent
was the result of light absorbance from the dyad thin films. Figure 6.11 shows a
photocurrent action spectra on the same graph as a spectra of the fraction of
photons absorbed for the same sample (measured with a UV-Vis
spectrophotometer). The photocurrent in the action spectra is the magnitude of
the measured current so that both the absorbance and the current appear on the
same side of the graph. Clearly, the maximum of the photocurrent is observed to
appear very close to the Soret of the dyad thin film. Furthermore, there are peaks
in the photocurrent where porphyrin Q-band absorption peaks occur (500-700
nm). This gives clear evidence that the photocurrent results are due to the dyad
thin film, and not some experimental artifact.
0
2 10-8
4 10-8
6 10-8
8 10-8
1 10-7
1.2 10-7
1.4 10-7
0
0.1
0.2
0.3
0.4
0.5
0.6
300 400 500 600 700 800
Dyad Film Photocurrent Action Spectra and Fraction of Photons Absorbed
Photocurrent
Optical Spectra
Pho
tocu
rrent
(A)
Fraction of Photons A
bsorbed by Film
Wavelength (nm)
Figure 6.11. Comparison of photocurrent action spectra of the dyad film with the spectra of the fraction of photons absorbed by the sample. The shape of the photocurrent action spectra (black) is very similar to the shape of the photon absorbance spectra. Photocurrent is typically negative, but is displayed positively here.
Data from almost all hexanethiol drop action spectra (10 out of 14)
showed measurable photocurrent, while only 5 out of 14 of the clean mercury
drops showed measurable photocurrent in the action spectra. In both the
hexanethiol drops and the bare mercury drops, there was a spot where action
spectra was not measured at all due to problems with the apparatus. The increase
incidence of measurable photocurrent with the hexanethiol drop coincides with
130
previous results showing increased white light photocurrent with the hexanethiol
drop compared to the bare mercury drop. Furthermore, the sign of the
photocurrent in the action spectra with the bare mercury drop was occasionally
positive, meaning that electrons went to the ITO as opposed to the mercury drop.
This was never seen with the hexanethiol drop. Figure 6.12 shows an action
spectra of a typical hexanethiol drop compared to the action spectra of a plain
mercury drop with current going in the positive direction. This may help explain
why the hexanethiol drop shows more overall efficiency. The mercury drop-
dyad-ITO geometry may not provide enough driving force for the electrons and
holes to move in a given direction with 100% certainty, meaning that slight
variations in the film may lead to a switching of the current signal due to the lack
of diode-like behavior from the
metals. However, when a
hexanethiol molecule is added,
this could provide the correct
dipole moment to control the
direction of the current, and
thus prevent recombination,
increasing photovoltaic
efficiencies.
The IV curves were
analyzed by examining the conductance of each curve from 0 V to 0.05 V.
Figure 6.12. Comparison of photocurrent action spectra. On the left, a typical hexanethiol mercury drop action spectra on dyad thin film shows with negative current. On the right is the action spectra for a bare mercury drop with current going in the positive direction.
131
Results showed a slight increase in conductivity upon illumination, with an
average increase of approximately 5 ± 3.7 % for hexanethiol drops and 2.1 ± 3.8
% for bare mercury drops. Increases in conductance upon illumination are
consistent with the STM results on single porphyrin-fullerene dyad on ITO, which
showed that a long lived charge separated state can occur after illumination of the
sample which most likely leads to increases in photo-conductance.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 0.02 0.04 0.06 0.08 0.1 0.12
0 0.02 0.04 0.06 0.08 0.1 0.12
Pow er Density Versus C ell S izeSpot 1Spot 2
Max
imum
Pos
sibl
e P
ower
Den
sity
(nW
/cm
2 )
Ce ll S ize (cm2)
F
Figure 6.13. Maximum possible power density versus solar cell size. The maximum possible power density is calculated as the product of the transient photovoltage and the transient photocurrent.
Finally, the potential maximum power density of the cells were compared
with the area of the cell (contact area of the mercury). Figure 6.13 shows a graph
of the potential power per cm2 (calculated as the product of the transient
photocurrent and the transient photovoltage) taken for two positions on the same
132
133
dyad thin film sample. This particular sample was not one of the six samples
originally made for the experiment, but was made later. For each position, the
area of the cell was gradually increased, taking a transient photocurrent and
photovoltage measurement before increasing the size of the mercury contact. In
general, as the cell size gets larger, the potential power density increases. This is
a unique application of the mercury drop electrode, as most solar cell area tests
have to be conducted by making several permanent metal contacts on an organic
film. Changes in photovoltaic efficiency, could then be due to changes in the
sample from one location to another. With the mercury drop, these tests were
able to be performed in the same general location on the organic sample. The
increase in maximum possible power density with increased cell size may be due
to better contact with the organic film as the mercury drop is pressed against the
sample.
6.3.3. Calculations of photocurrent efficiencies
Although the action spectra and transient photocurrent measurements
show that photocurrent is produced from the dyad thin films upon illumination, it
is difficult to tell from these measurements how efficient the cell is at converting
light to current. In order to determine this efficiency, the incident photon-to-
electron conversion efficiency (IPCE) was calculated. This quantifies the
percentage of photons that are converted into electrons during the illumination of
a solar cell. The calculation of the IPCE is based upon the wavelength of the light
and the incident illumination intensity which was recorded on the photodiode.
The calculation to determine IPCE was
λDD
Ds
AIRAI
IPCE *1240= 6.6
Where Is and ID are the current at the sample and the photodiode at wavelength, λ
(in nanometers), R is the responsivity of the photodiode at the specified
wavelength, and AS and AD are the area of the mercury drop contact and the
photodiode. 1240 is a conversion factor that relates the responsivity of the
photodiode to its quantum efficiency. It is the wavelength associated with 1 eV of
photon energy.
After inserting current data into this equation, a spectra of IPCE was
developed for every action spectra. An example of an IPCE spectra is shown in
Figure 6.14. The highest IPCE value occurs around the Soret region. For all
measurable IPCE spectra for both the bare mercury drop and the hexanethiol
drop, the wavelength of maximum IPCE was 447 ± 6.2 nm. Of the six samples
and 30 spots tested, the maximum IPCE was about 0.5% in the Soret region.
Reported values for the IPCE for bulk heterojunction solar cells incorporating
fullerene derivatives have been over 60% in some spectral regions. The IPCE
values for these self assembled films is much lower in comparison, but this may
still be due to the similar work functions of the ITO and mercury drop.
134
0
0 . 1
0 . 2
0 . 3
0 . 4
0 . 5
0 . 6
4 0 0 5 0 0 6 0 0 7 0 0 8 0 0
I n c i d e n t P h o t o n - t o - E l e c t r o n C o n v e r s i o n E f f i c i e n t y
Effi
cien
cy (%
)
W a v e le n g t h ( n m )
Figure 6.14. Incident photon-to-electron conversion efficiency for dyad thin film and mercury drop electrode.
6.4. Conclusions
The spontaneous formation of the dyad thin films on the ITO after
porphyrin protonation is an intriguing feature of these molecules. There is
currently quite an ongoing search for suitably cheap and efficient organic
materials that can be utilized as active photovoltaic elements, and the potential to
have a material assemble on a surface in a beneficial way would be an important
breakthrough in organic photovoltaics. For this reason, instrumentation to
examine photovoltaic materials is a vital component in understanding the
potential benefits of a new material.
There is, unfortunately, very little to say conclusively about the
organization of the dyad film on the ITO. Surface characterization techniques are
notoriously difficult on ITO, and prove even harder with very thin organic films.
135
136
However, the fact that the film only grows in the protonated state indicates that
the trifluoroacetate ions are actually a component of the film, as the fully
protonated molecule, where the porphyrin is protonated in addition to the pyridine
attached to the porphyrin and the amine on the fullerene, would not be prone to
aggregating without the presence of counter ions. It may be that the
trifluoroacetate ions form a layer on the ITO first, followed by the dyad
molecules, and the two species of molecules continue to alternate to form the
film. The photovoltaic experiments do not indicate a very ordered film, or
photovoltage would be higher due to an accumulated dipole moment.
It is clear from both the AFM results and the calculations of photocurrent
efficiencies that the conducting AFM is not the ideal instrument for studying
organic photovoltaic materials. The difficulties in working with an AFM
cantilever in an illuminating environment are plentiful, from the deflection of the
cantilever, to the potential of creating p-n junction devices between the substrate
and the tip. Furthermore, based upon photocurrent calculations, the size of the
AFM tip is not adequately large enough to be able to detect sizeable photocurrent.
It could be possible that with better electronics, the signal to noise ratio could be
reduced and the photocurrent detected, but this is beyond the current capabilities
of the illuminated SPM system.
The mercury drop instrument, however, proved to be much more suited to
determining photovoltaic efficiencies in these dyad films. The size of the
mercury contact was much larger than the size of the AFM contact, allowing for
137
more photocurrent to be obtained. The utilization of a lamp and a monochromator
also allowed for the sample to be illuminated in both white light and at very
specific wavelengths so the properties of the film could be examined in a spectral
manner.
From the mercury drop experiments, it was determined that while the
films showed some photovoltaic properties, the photovoltages and overall power
were very low. An important factor in achieving reliable photocurrent
measurements was the use of a monolayer on the mercury drop. It is believed that
this monolayer helps direct current in a specific direction relative to the two
electrodes, thus preventing recombination at the mercury interface which could
lead to losses in photocurrent efficiency. Despite the increased performance with
the hexanethiol, however, the IPCE values were still comparatively low, but this
may also be partially due to the similar electrode work functions. Regardless of
the low IPCE, though, the films still exhibit novel assembly features with
measurable, if not phenomenal, photovoltaic properties.
138
REFERENCES
1 G. Binnig and H. Rohrer, Helvetica Physica Acta 55 (6), 726 (1982).
2 B. L. Zhang and E. K. Wang, Electrochimica Acta 39 (1), 103 (1994).
3 S. Kerfriden, A. H. Nahle, S. A. Campbell et al., Electrochimica Acta 43 (12-13), 1939 (1998).
4 S.M. Lindsay, Introduction to Nanoscience. (Oxford University Press, Oxford, 2010).
5 G. Binnig, C. F. Quate, and C. Gerber, Physical Review Letters 56 (9), 930 (1986).
6 W. H. Han and S. M. Lindsay, Applied Physics Letters 72 (13), 1656 (1998).
7 A. Ebner, F. Kienberger, G. Kada et al., Chemphyschem 6 (5), 897 (2005).
8 L. Lin, H. D. Wang, Y. Liu et al., Biophysical Journal 90 (11), 4236 (2006).
9 T. R. Albrecht, P. Grutter, D. Horne et al., Journal of Applied Physics 69 (2), 668 (1991).
10 Sergei Magonov and John Alexander.
11 T. Trenkler, T. Hantschel, R. Stephenson et al., Journal of Vacuum Science & Technology B 18 (1), 418 (2000).
12 Neil W. Ashcroft and N. David Mermin, Solid State Physics. (Thompson Learning, 1976).
13 A. Aviram and M. A. Ratner, Chemical Physics Letters 29 (2), 277 (1974).
139
14 P. Hohenberg and W. Kohn, Physical Review B 136 (3B), B864 (1964).
15 J. Tomfohr and O. F. Sankey, Journal of Chemical Physics 120 (3), 1542 (2004).
16 F. Evers, F. Weigend, and M. Koentopp, Physical Review B 69 (23) (2004).
17 R. A. Marcus, Journal of Chemical Physics 24 (5), 966 (1956).
18 K. W. Hipps, Science 294 (5542), 536 (2001).
19 H. Park, J. Park, A. K. L. Lim et al., Nature 407 (6800), 57 (2000).
20 W. J. Liang, M. P. Shores, M. Bockrath et al., Nature 417 (6890), 725 (2002).
21 T. Morita and S. Lindsay, Journal of the American Chemical Society 129 (23), 7262 (2007).
22 X. D. Cui, A. Primak, X. Zarate et al., Science 294 (5542), 571 (2001).
23 A. M. Rawlett, T. J. Hopson, L. A. Nagahara et al., Applied Physics Letters 81 (16), 3043 (2002).
24 G. K. Ramachandran, J. K. Tomfohr, J. Li et al., Journal of Physical Chemistry B 107, 6162 (2003).
25 B. Q. Xu and N. J. J. Tao, Science 301 (5637), 1221 (2003).
26 J. He, O. Sankey, M. Lee et al., Faraday Discussions 131, 145 (2006).
27 F. Chen, Z. F. Huang, and N. J. Tao, Applied Physics Letters 91 (16) (2007).
140
28 B. Q. Q. Xu, X. L. L. Li, X. Y. Y. Xiao et al., Nano Letters 5 (7), 1491 (2005).
29 B. Q. Xu, X. Y. Xiao, and N. J. Tao, Journal of the American Chemical Society 125 (52), 16164 (2003).
30 Iris Visoly-Fisher, Kayvon Daie, Yuichi Terazono et al., PNAS 103 (23) (2006).
31 X. Y. Xiao, L. A. Nagahara, A. M. Rawlett et al., Journal of the American Chemical Society 127 (25), 9235 (2005).
32 X. Y. Xiao, B. Q. Xu, and N. J. Tao, Nano Letters 4, 267 (2004).
33 H. Shpaisman, E. Salomon, G. Nesher et al., Journal of Physical Chemistry C 113 (8), 3313 (2009).
34 A. Salomon, R. Arad-Yellin, A. Shanzer et al., Journal of the American Chemical Society 126, 11648 (2004).
35 The National Institutes of Health National Human Genome Research Institute, NHGRI Seeks Next Generation of Sequencing Technologies, (2004), http://www.genome.gov/12513210.
36 F. Sanger, S. Nicklen, and A. R. Coulson, Proceedings of the National Academy of Sciences of the United States of America 74 (12), 5463 (1977).
37 Paula Yurnakis Bruice, Organic Chemistry, 5th ed. (Pearson Prentice Hall, Upper Saddle River, 2007).
38 J. Shendure and H. L. Ji, Nature Biotechnology 26 (10), 1135 (2008).
39 D. Branton, D. W. Deamer, A. Marziali et al., Nature Biotechnology 26 (10), 1146 (2008).
141
40 J. J. Kasianowicz, E. Brandin, D. Branton et al., Proceedings of the National Academy of Sciences of the United States of America 93 (24), 13770 (1996).
41 M. Akeson, D. Branton, J. J. Kasianowicz et al., Biophysical Journal 77 (6), 3227 (1999).
42 A. Meller, L. Nivon, E. Brandin et al., Proceedings of the National Academy of Sciences of the United States of America 97 (3), 1079 (2000).
43 Author, The Lindsay Lab: Research Electronic Sequencing by Recognition, http://www.biodesign.asu.edu/labs/lindsay/research.
44 T. Ohshiro and Y. Umezawa, Proceedings of the National Academy of Sciences of the United States of America 103 (1), 10 (2006).
45 J. He, L. S. Lin, H. Liu et al., Nanotechnology 20 (7) (2009).
46 J. He, L. S. Lin, P. M. Zhang et al., Nano Letters 8 (8), 2530 (2008).
47 S. Chang, J. He, A. Kibel et al., Nature Nanotechnology 4 (5), 297 (2009).
48 Jeffrey L. Hutter and John Bechhoefer, Rev. Sci. Instrum. 64 (7) (1993).
49 Jeffrey L. Hutter, Calibration of AFM Cantilevers, (2005), http://www.physics.uwo.ca/~hutter/calibration/afmcal.html.
50 Jacob N. Israelachvili, Intermolecular and Surface Forces, 2 ed. (Elsevier Ltd., London, 1985).
51 J. Sponer, P. Jurecka, and P. Hobza, Journal of the American Chemical Society 126 (32), 10142 (2004).
52 J. M. Williams, T. J. Han, and T. P. Beebe, Langmuir 12 (5), 1291 (1996).
142
53 U.S. Department of Energy, 2008 Solar Technologies Market Report, January, 2010,
54 U.S. Department of Energy, (June 16, 2010).
55 Thomas Surek and The National Renewable Energy Laboratory, PROGRESS IN U.S. PHOTOVOLTAICS: LOOKING BACK 30 YEARS AND LOOKING AHEAD 20, http://www.nrel.gov/pv/thin_film/docs/surek_osaka_talk_final_vgs.pdf
56 Y. Kim, S. Cook, S. M. Tuladhar et al., Nature Materials 5 (3), 197
(2006).
57 J. J. M. Halls, K. Pichler, R. H. Friend et al., Applied Physics Letters 68 (22), 3120 (1996).
58 A. Haugeneder, M. Neges, C. Kallinger et al., Physical Review B 59 (23), 15346 (1999).
59 L. A. A. Pettersson, L. S. Roman, and O. Inganas, Journal of Applied Physics 86 (1), 487 (1999).
60 T. J. Savenije, J. M. Warman, and A. Goossens, Chemical Physics Letters 287 (1-2), 148 (1998).
61 Brian O'Regan and Michael Gratzel, Nature 353, 737 (1991).
62 M. Gratzel, Journal of Photochemistry and Photobiology a-Chemistry 164 (1-3), 3 (2004).
63 P. Wang, S. M. Zakeeruddin, J. E. Moser et al., Nature Materials 2 (6), 402 (2003).
64 G. Yu, J. Gao, J. C. Hummelen et al., Science 270 (5243), 1789 (1995).
143
65 F. Padinger, R. S. Rittberger, and N. S. Sariciftci, Advanced Functional Materials 13 (1), 85 (2003).
66 G. Li, V. Shrotriya, J. S. Huang et al., Nature Materials 4 (11), 864 (2005).
67 Jin Young Kim, Kwanghe Lee, Nelson E. Coates et al., Science 317, 222 (2007).
68 H. Hoppe and N. S. Sariciftci, Journal of Materials Chemistry 16 (1), 45 (2006).
69 W. U. Huynh, J. J. Dittmer, and A. P. Alivisatos, Science 295 (5564), 2425 (2002).
70 M. Law, L. E. Greene, J. C. Johnson et al., Nature Materials 4 (6), 455 (2005).
71 Y. M. Kang, N. G. Park, and D. Kim, Applied Physics Letters 86 (11) (2005).
72 S. Kuhn, U. Hakanson, L. Rogobete et al., Physical Review Letters 97 (1) (2006).
73 F. Tam, G. P. Goodrich, B. R. Johnson et al., Nano Letters 7 (2), 496 (2007).
74 B. P. Rand, P. Peumans, and S. R. Forrest, Journal of Applied Physics 96 (12), 7519 (2004).
75 M. Westphalen, U. Kreibig, J. Rostalski et al., Solar Energy Materials and Solar Cells 61 (1), 97 (2000).
76 D. Kuciauskas, S. Lin, G. R. Seely et al., Journal of Physical Chemistry 100 (39), 15926 (1996).
144
77 D. Kuciauskas, P. A. Liddell, S. Lin et al., Journal of Physical Chemistry B 104 (18), 4307 (2000).
78 F. Chen, C. Nuckolls, and S. Lindsay, Chemical Physics 324 (1), 236 (2006).
79 S. W. Hahm, H. S. Hwang, D. Kim et al., Electronic Materials Letters 5 (4), 157 (2009).
80 M. Y. Chan, C. S. Lee, S. L. Lai et al., Journal of Applied Physics 100 (9) (2006).
81 D. A. Rider, K. D. Harris, D. Wang et al., Acs Applied Materials & Interfaces 1 (2), 279 (2009).
82 H. B. Michaelson, Journal of Applied Physics 48 (11), 4729 (1977).
83 Y. Park, V. Choong, Y. Gao et al., Applied Physics Letters 68 (19), 2699 (1996).
84 K. Sugiyama, H. Ishii, Y. Ouchi et al., Journal of Applied Physics 87 (1), 295 (2000).
85 Y. J. Liu and H. Z. Yu, Chemphyschem 3 (9), 799 (2002).
APPENDIX A
LABVIEW CODES
145
All labview codes were written in either Labview 8.2 or Labview 8.5.
This appendix covers most of the major programs used in this dissertation work,
and includes commented diagrams for the main portions of each code.
Main programming template
Complex labview code can be difficult to write. For this reason, a
programming template was utilized for most of the codes. This template allowed
for the program to remain idle unless a specific panel was called by a control on
the front panel (user interface). The block diagram for this template is shown
below.
146
The bottom structure is the event structure which is linked to different controls on
the front panel. When one of those controls is changed by the user, the event
structure identifies the control and passes a corresponding string up into a queue
structure. The string then goes through the queue, where it is de-queued and
passed to the main case structure which is held in a while loop. The element then
determines which case/state to call. Then, whatever functions are in that state are
completed and a new string is put into the queue, directing the program as to
which case/state to call next. In the event that there are no further functions to
complete, the queue calls an [Idle] case, which puts nothing new into the queue,
idling the program until another case is called by the user.
To the left of the program is a data cluster where values can be held. A
connection to this cluster is run through all cases/states in the case structure so
that data values can be called, manipulated and changed within a case. This is an
efficient way of storing data in Labview, and allows for “cleaner” programming
code. Data is added to the cluster by simply adding constants to it and naming
those constants appropriately to describe what they are in the program. Again,
this programming template was utilized for most labview codes.
Base-nucleoside tip withdraw and approach program
The main author of the withdraw and approach program was Shuai Chang.
The most relevant cases are included here. The program allows the STM or AFM
tip to be controlled at the surface under a setpoint current value. Then, the user
147
can decide when to withdraw and re-approach the surface. The tip is held at the
surface in the programming case shown below.
The above program takes a sample of the current and runs it through a PID VI
which calculates a piezo voltage correction based upon the setpoint value and the
PID values, which are determined by the user. If this correction value is very
different than the current piezo voltage, then a smaller difference is written to the
piezo controller (which controls the height of the tip). If the difference is small,
then the corrected voltage is directly written to the piezo.
Once the user decides to withdraw the tip, the PID loop is broken and the
program moves to a new case/state in order to withdraw and re-approach to the
surface. That case is shown below.
148
In the above panel, the withdraw time, withdraw distance and scanner
sensitivity are used to calculate values (frequency and amplitude) for a triangle
waveform to be written to the PZT controller in order to withdraw and approach
to the surface. The value for the piezo voltage is obtained and used as a baseline
value for the triangle waveform. Once the waveform is written, the program
eventually returns to PID control.
AFM tip calibration program
The AFM tip calibration program first acquires a set of data for cantilever
deflection at very high sampling rates for a short period of time. Typically, the
149
program acquires 50,000 samples at a rate of 800,000 samples per second. The
program then calculates the power spectral density of the data. This
sampling/calculating procedure repeats for a set number of times (typically 150),
while compiling the power spectral densities. The diagrams for the cases to
acquire the data and calculate the power spectral densities are below.
150
Once the number of acquisitions has been reached, the program then goes
on to average all of the power spectral densities that were calculated. The
“Average PSD” case repeats for the number of power spectral densities that were
calculated. Every time the case is called, it extracts one density value from each
of the compiled arrays such that the density values it extracts correspond to the
same frequency from each array. It then averages all of the densities at that
frequency and, over time, compiles an average power spectral density array
giving the average power spectral density at each frequency. In this way, all of
the power spectral densities are averaged together. The average power spectral
density is then plotted and the program idles. The case for this averaging
procedure is shown below.
Once the average power spectral density is plotted, the user can search for
the resonant peak and input into the program the start and end frequencies for the
151
peak. The program then can reduce the average power spectral density array to
just this peak, and then curve fit the peak to a Lorentzian function. Below shows
the case to reduce the power spectral density array to the user specified limits and
the case to curve fit the peak to a Lorentzian function. Note that the panel to
reduce the power spectral density also numerically calculates the area under the
curve for the array in that range. This was done as another way to get the area
under the curve, but was not typically utilized in final tip calibrations.
152
Another case calculates the area under the curve from the curve fit
coefficients determined by the curve fit VI. Finally, the user can hit a button to
calculate the spring constant. The case that calculates the spring constant from
equation 3.1 is below.
153
CAFM base-nucleoside data analysis program
The data analysis program for the CAFM DNA base-nucleoside
experiment involves multiple steps. The two most important calculations involve
calculating the adhesion force of the retraction and calculating the unstretched
(F=0) current value. The case that calculates the adhesion force also converts all
the deflection values into force values. The user inputs the spring constant and
the piezo sensitivity. Another case calculates the retraction distance from the
voltage on the piezo and the piezo sensitivity. Once the distance has been
calculated the adhesion force can be calculated. The user identifies two points in
the deflection contact region (straight line) between which the program finds the
slope of the deflection-distance curve. This slope gives the deflection sensitivity
of the cantilever in terms of V/nm. The minimum force value is also obtained
either manually or automatically, and the deflection at F=0 is input manually by
the user. The difference between the deflection at F=0 and the minimum
deflection gives the adhesion deflection. This is then converted to force by
dividing by the sensitivity to obtain a distance in nanometers and then multiplying
by the spring constant to get the force. A similar process was completed on the
entire deflection array to convert all the deflection data into force values, which
was plotted against distance. The case that calculates all of this is shown below.
154
To get the value of the current when F=0, the force array was analyzed to
determine the index of the F=0 value. Once this index was determined, it was
utilized to extract the current value from the current array. The case that
calculates the unstretched current is shown below.
155
STM break junction experiment
The program to control the tip for the STM break junction experiment has
two main components. One case of the program holds the tip under PID control.
That case is shown below.
After running the PID loop for a set number of times, the program
automatically goes to a case that withdraws and approaches the tip a set distance
given a set time. The case that does this is shown below. Like the DNA
nucleoside-base program, it utilizes a triangle wave function to produce the
withdraw and approach of the STM tip through the scanner piezo voltage and then
returns the tip to PID control.
156
STM break junction data recording
The current withdraw curves, along with the shutter voltage, were
recorded on a Tektronix oscilloscope and transferred through a labview program
to an open binary file. Another text file recorded the number of points per curve
along with the time increment between data points. The oscilloscope came with a
labview program that was able to control and record directly from the
oscilloscope. The signals were then decomposed and saved into the binary file.
The case for this data transfer process is shown below.
157
STM break junction “step” analysis
The STM break junction retraction curves were analyzed in order to
determine where the current “steps” occurred. A histogram is first made for each
curve. The user determines the length of a window to slide over each histogram
and the program determines the maximum and minimum values in that window.
If the difference between maximum and minimum is above a user controlled
threshold, and if the maximum occurs somewhere in the middle of the window
(not first or last value), then the current at the maximum is assigned a value of 1.
This procedure is applied as the window is slid up the histogram array. The case
that creates the single curve histogram and identifies “steps” is shown below.
158
If the user decides to add the steps from a curve to a histogram he/she can
tell the program to add the curve. The program then determines whether or not
the illuminating laser was on during the particular tip retraction (determined from
the voltage on the optical shutter) and then either adds it to the light-on histogram,
or adds it to the light-off histogram. The case to add to the compiled light-off
histogram is shown below.
159
Acquire data for CAFM of dyad thin film program
The program to acquire data from the CAFM to study photocurrent effects
in a dyad thin film was relatively simple. The program set up the number of
channels to acquire and then acquired data for four channels at a sampling rate set
by the user for a given time interval. The interval would repeat and that data
would be compiled until the while loop that encompassed the acquisition VI was
stopped. The case that acquires the data is below.
CAFM of dyad thin films data viewing program
The program to view the data acquired from the CAFM of dyad thin films
was also relatively straightforward. The file where the data was saved was open
160
and the array was indexed in order to decompose the different signals. These
signals were then graphed against a time axis. These signals changed depending
upon what was connected to the data acquisition device, but the first signal was
always current. The case that graphs the data is shown below.
Mercury contact area analysis program
A picture was taken for every mercury contact in the mercury drop
electrode experiment on the dyad thin films. A labview program was created in
order to determine the size of the contact area. The program takes a picture file
and converts it to an intensity graph. There are three cursors on the intensity
graph that the user can adjust. The program takes the positions of these three
cursors on the intensity graph and fits a circle to those three points, giving the
radius of the circle in pixels. For every new mercury drop (not every new
contact), the radius of the precision screw was determined through this method
161
and compared with the actual radius of the precision screw, measured with a
micrometer. This radius could be used later to determine actual distances in the
image Then pixel radius of every mercury contact was determined by placing the
cursors around the edge of the mercury contact circle. The case that created the
intensity graph and cursors and the case that calculated the radius of the circle are
shown below.
162
Mercury electrode results analysis program
The data from the mercury drop electrode apparatus was analyzed with
another labview program. The user indicates a sample number and a spot number
in the front panel of the program. The experimental data was saved in a particular
way in order to recognize sample and spot numbers. For example, the
photocurrent of sample 3 spot 2 would be saved as s3s2pc.txt. The program then
used the sample number and spot number to open all files associated with that
sample and spot number. The case that opens the file is shown below.
The different measurements were then analyzed. First, the photocurrent
was analyzed by determining the difference in current before and after the light
was turned on. In the same panel the average dark current was also determined,
along with the standard deviation of the dark current. The case that analyzed the
photocurrent is shown below. The photovoltage was similarly analyzed.
163
The action spectra was analyzed in a separate case to determine the peak
photocurrent and the wavelength of the peak photocurrent. In another case, the
IV curves were analyzed by determining the current values from 0 to 0.05 V. The
conductivity was found by taking a linear fit to the values. Both the dark
conductivity and the light conductivity were determined in this way. In the same
case, the responsivity data for the photodiode, which was saved in a separate text
file, was converted into a numeric array. This case is shown below.
164
Finally, the IPCE was determined from the action spectra of the sample,
the action spectra of the photodiode (opened and transferred to an array in another
case) and the responsivity data. The responsivity values are given for every 10
nm of optical wavelength, whereas the action spectra values are every 5 nm. So
values for the responsivity were interpolated to create an array of responsivity
values for every 5 nm. Using equation 6.6, the IPCE is calculated in the case
below.
165